《3453 PowerVM AIX Monitoring SAN IO to from memory V3.7 Sept15-2024 earlj.pdf》由會員分享，可在線閱讀，更多相關《3453 PowerVM AIX Monitoring SAN IO to from memory V3.7 Sept15-2024 earlj.pdf（213頁珍藏版）》請在三個皮匠報告上搜索。
1、October 21-24,2024Mandalay Bay Convention CenterLas Vegas,NevadaPowerVM AIX:Monitoring SAN I/O to/from memory Session Code 3453Earl Jew ()310-251-2907 cell Consulting IT Management ConsultantIBM Power and IBM Systems Storage IBM Technology Expert Labs-Power delivery practiceIBM Technology Expert Lab
2、s/USA Copyright IBM Corporation 20242PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memo
3、ry before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 50%idle Problematic Themea999007 PowerVM AIX:The 6
4、layers of the PowerVM CPU nomenclature at SMT8PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Minimizing CPUtime recycling at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memoryEvaluating SAN I/Oa999011 P
5、owerVM AIX:Mapping the path of workload executionDeep Science monitoringPowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN
6、 Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 1.0ms average se
7、rvice times in todays OEM systems storage technologies,i.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This session introduces the tactics for:Monitoring SAN I/O to/from memory34523453IBM Technology Expert Labs/USA Copyright IBM
8、Corporation 20244PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing t
9、o SAN,i.e.,real memory is the other endCompared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage technologies,i
10、.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This session introduces the tactics for:Monitoring SAN I/O to/from memory All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushi
11、ng to SAN IBM Technology Expert Labs/USA Copyright IBM Corporation 20245PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN
12、 Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 1.0ms average se
13、rvice times in todays OEM systems storage technologies,i.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This session introduces the tactics for:Monitoring SAN I/O to/from memory All SAN Reads must be received in real memory and al
14、l SAN Writes must be staged in real memory before flushing to SAN but how quickly memory receives SAN Reads and stages SAN writes is ignored.IBM Technology Expert Labs/USA Copyright IBM Corporation 20246PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iost
15、at is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN
16、Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage technologies,i.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in perform
17、ance.This session introduces the tactics for:Monitoring SAN I/O to/from memory All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN but how quickly memory receives SAN Reads and stages SAN writes is ignored.The immediate readiness of m
18、emory to support SAN I/O is an unseen emerging factor in performance.IBM Technology Expert Labs/USA Copyright IBM Corporation 20247PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forge
19、t:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughpu
20、t and responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage technologies,i.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This session introduces the tactics for:Monitoring SAN I/O to/from
21、memoryAll SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN but how quickly memory receives SAN Reads and stages SAN writes is ignored.The immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This
22、 session introduces the tactics for:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 20248PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forge
23、t:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughpu
24、t and responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage technologies,i.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This session introduces the tactics for:Monitoring SAN I/O to/from
25、memoryEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinalResponsiveness of ThroughputHow%time_a
26、ctive affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineIBM Technology Expert Labs/USA Copyright IBM Corporation 20249PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct
27、practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN
28、writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage technologies,i.e.,IBM FlashSystems,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.This session int
29、roduces the tactics for:Monitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 202410PowerVM AIX:Monitoring SAN I/O to/from memoryMonitor
30、ing SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinalResponsiveness of ThroughputHow%t
31、ime_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology
32、 Expert Labs/USA Copyright IBM Corporation 202411PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of s
33、ystems storage at saturation How%time_active affects the latitudinal Responsiveness of Throughput How%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at s
34、aturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 202412SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat+iostat-DRTl 60 10IBM Te
35、chnology Expert Labs/USA Copyright IBM Corporation 202413SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatRead ResponsivenessWrite ResponsivenessService Queue when fullphysical I/O Throughput+iostat-DRTl 60 10SAN I/O performa
36、nce=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202414SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_through
37、put_mode=2Evaluating SAN I/O performance with AIX: of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)IBM Technology Expert Labs/USA Copyright IBM Corporation 202415SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O per
38、formance with AIX:iostat%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.IBM Technology Expert Labs/USA Copyright IBM Corporation 202416SPLPAR ventura:Power10 E108
39、0 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.Objective:Demonst
40、rating how%time_active affects SAN I/O responsiveness-of-throughput IBM Technology Expert Labs/USA Copyright IBM Corporation 202417SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physical disk/tape over t
41、he 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.Objective:Demonstrating how%time_active affects SAN I/O responsiveness-of-throughputbut first,we must define Throughput and Responsiveness in AIX:iostat-DRTl 6
42、0 10 output.IBM Technology Expert Labs/USA Copyright IBM Corporation 202418SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)%t
43、ime_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.Objective:Demonstrating how%time_active affects SAN I/O responsiveness-of-throughputbut first,we must define Throughput and Responsiveness in AIX:iostat-DRTl 60 10 output.IBM Technology Expert Labs/USA Copyright IBM
44、 Corporation 202419SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX: of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)IBM Technology Expert Labs/USA Copyright IBM Corporation 202420SPLPAR ventur
45、a:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O Throughput over the 60sec interval(default is in bytes per second) Technology Expert Labs/USA Copyright IBM Corporation 202421SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00
46、 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical transfers per sec of indeterminate size(aka IOPS)over the 60sec Technology Expert Labs/USA Copyright IBM Corporation 202422SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I
47、/O performance with AIX:iostatphysical Read Throughput over the 60sec interval(default is in bytes per second) Technology Expert Labs/USA Copyright IBM Corporation 202423SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysic
48、al Write Throughput over the 60sec interval(default is in bytes per second) Technology Expert Labs/USA Copyright IBM Corporation 202424SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatThroughput is the amount of data transfer
49、red per second(bps)of the the number of transfers per second(tps)over the 60sec intervalphysical I/O T Technology Expert Labs/USA Copyright IBM Corporation 202425SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical Read
50、Responsiveness over the 60sec interval(milliseconds) ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202426SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatRead Responsiveness is the service times of re
51、ad transfers of the read transfers per second(rps)over the 60sec interval in ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202427SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX: Write Responsivenessover th
52、e 60sec interval(milliseconds)Write ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202428SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX: ResponsivenessWrite Responsiveness is the service timesof write tran
53、sfers of the write transfers persecond(wps)over the 60sec interval in millisecondsIBM Technology Expert Labs/USA Copyright IBM Corporation 202429SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatWrite ResponsivenessRead Respon
54、sivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202430SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatWrite ResponsivenessRead ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202431SPL
55、PAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatWrite ResponsivenessRead ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202432SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=
56、2Evaluating SAN I/O performance with AIX:iostatWrite ResponsivenessRead ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202433SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatResponsiveness is the servi
57、ce times of transfers of the xfers per sec=(rps+wps)over the 60sec interval in milliseconds+=+ResponsivenessWrite ResponsivenessRead ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corporation 202434SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating S
58、AN I/O performance with AIX: 60 10Avg ServQ sizeIBM Technology Expert Labs/USA Copyright IBM Corporation 202435SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX: 60 10Avg ServQ sizeIBM Technology Expert Labs/USA Copyright IBM Corpor
59、ation 202436SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX: Queue when fullIBM Technology Expert Labs/USA Copyright IBM Corporation 202437SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SA
60、N I/O performance with AIX:iostatservice queue full=1.0across multiple 60sec intervals begins degrading Responsivenessservice queue full=10.0across multiple 60sec intervals begins degrading Throughputservice queue full=100.0 across multiple 60sec intervals is a critical performance Queue when fullIB
61、M Technology Expert Labs/USA Copyright IBM Corporation 202438SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatmilliseconds of latency only when the Service Queue is fullService Queue when Technology Expert Labs/USA Copyright
62、IBM Corporation 202439SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatRead ResponsivenessService Queue when fullphysical I/O Throughput+iostat-DRTl 60 10Write ResponsivenessIBM Technology Expert Labs/USA Copyright IBM Corpor
63、ation 202440SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat+iostat-DRTl 60 10physical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullThroughput is the amount of data transferred per second(bps)o
64、f the number of transfers per second(tps)over the 60sec intervalResponsiveness is the service times of transfers of the transfers per sec=(rps+wps)over the 60sec interval in millisecondsRead Throughput is the read data transferred(bread)of the read transfers per second(rps)over the 60sec interval in
65、 msecs Write Throughput is the write data transferred(bwrtn)of the write transfers per second(wps)over the 60sec interval in msecsRead Responsiveness is the service times of read transfers of the read transfers per second(rps)over the 60sec interval in msecs Write Responsiveness is the service times
66、 of write transfers of the write transfers per second(wps)over the 60sec interval in msecsIBM Technology Expert Labs/USA Copyright IBM Corporation 202441Evaluating SAN I/O performance with AIX:iostatRead ResponsivenessService Queue when fullphysical I/O ThroughputWrite ResponsivenessThroughput is th
67、e amount of data transferred per second(bps)of the number of transfers per second(tps)over the 60sec intervalResponsiveness is the service times of transfers of the transfers per sec=(rps+wps)over the 60sec interval in millisecondsRead Throughput is the read data transferred(bread)of the read transf
68、ers per second(rps)over the 60sec interval in msecs Write Throughput is the write data transferred(bwrtn)of the write transfers per second(wps)over the 60sec interval in msecsRead Responsiveness is the service times of read transfers of the read transfers per second(rps)over the 60sec interval in ms
69、ecs Write Responsiveness is the service times of write transfers of the write transfers per second(wps)over the 60sec interval in msecsIBM Technology Expert Labs/USA Copyright IBM Corporation 202442Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the Responsivenessof ThroughputSAN I
70、/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnThroughput is the amount of data transferred per second(bps)of the number of transfers per second(tps)over the 60sec int
71、ervalResponsiveness is the service times of transfers of the transfers per sec=(rps+wps)over the 60sec interval in millisecondsRead Throughput is the read data transferred(bread)of the read transfers per second(rps)over the 60sec interval in msecs Write Throughput is the write data transferred(bwrtn
72、)of the write transfers per second(wps)over the 60sec interval in msecsRead Responsiveness is the service times of read transfers of the read transfers per second(rps)over the 60sec interval in msecs Write Responsiveness is the service times of write transfers of the write transfers per second(wps)o
73、ver the 60sec interval in msecsIBM Technology Expert Labs/USA Copyright IBM Corporation 202443Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the Responsivenessof ThroughputThroughput is the amount of data transferred per second(bps)of the number of transfers per second(tps)over th
74、e 60sec intervalResponsiveness is the service times of transfers of the transfers per sec=(rps+wps)over the 60sec interval in millisecondsRead Throughput is the read data transferred(bread)of the read transfers per second(rps)over the 60sec interval in msecs Write Throughput is the write data transf
75、erred(bwrtn)of the write transfers per second(wps)over the 60sec interval in msecsRead Responsiveness is the service times of read transfers of the read transfers per second(rps)over the 60sec interval in msecs Write Responsiveness is the service times of write transfers of the write transfers per s
76、econd(wps)over the 60sec interval in msecsIBM Technology Expert Labs/USA Copyright IBM Corporation 202444Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per s
77、econd(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnThroughput is the amount of data transferred per second(bps)of the number of transfers per second(tps)over the 60sec intervalResponsiveness is the service times of transfers of the transfers per sec=(rps+wps)over
78、the 60sec interval in millisecondsRead Throughput is the read data transferred(bread)of the read transfers per second(rps)over the 60sec interval in msecs Write Throughput is the write data transferred(bwrtn)of the write transfers per second(wps)over the 60sec interval in msecsRead Responsiveness is
79、 the service times of read transfers of the read transfers per second(rps)over the 60sec interval in msecs Write Responsiveness is the service times of write transfers of the write transfers per second(wps)over the 60sec interval in msecsIBM Technology Expert Labs/USA Copyright IBM Corporation 20244
80、5SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat+iostat-DRTl 60 10Throughput is the amount of data transferred per second(bps)of the number of transfers per second(tps)over the 60sec intervalResponsiveness is the service ti
81、mes of transfers of the transfers per sec=(rps+wps)over the 60sec interval in millisecondsRead Throughput is the read data transferred(bread)of the read transfers per second(rps)over the 60sec interval in msecs Write Throughput is the write data transferred(bwrtn)of the write transfers per second(wp
82、s)over the 60sec interval in msecsRead Responsiveness is the service times of read transfers of the read transfers per second(rps)over the 60sec interval in msecs Write Responsiveness is the service times of write transfers of the write transfers per second(wps)over the 60sec interval in msecsSAN I/
83、O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202446SPLPAR
84、ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iost
85、at.Objective:Demonstrating how%time_active affects SAN I/O responsiveness-of-throughputbut first,we must define Throughput and Responsiveness in AIX:iostat-DRTl 60 10 output.IBM Technology Expert Labs/USA Copyright IBM Corporation 202447PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/O
86、Evaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinal Responsiveness of Throughput How%time_activ
87、e affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert L
88、abs/USA Copyright IBM Corporation 202448PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems st
89、orage at saturation How%time_active affects the latitudinalResponsiveness of ThroughputHow%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation t
90、o/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 202449SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physical disk/t
91、ape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.SAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per s
92、econd(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202450SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physic
93、al disk/tape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.SAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transfer
94、red per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202451SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm
95、_throughput_mode=2Evaluating SAN I/O performance with AIX:iostat%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.Objective:Demonstrating how%time_active affects th
96、e Responsiveness of Throughput.SAN I/O performance=the Responsiveness of Throughput SAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA
97、Copyright IBM Corporation 202452SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the Responsiveness of Throughput SAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferre
98、d per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objective:Demonstrating how%time_
99、active affects the longitudinal Responsiveness of Throughput.%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)%time_active is an overlooked factor in Evaluating SAN I/O performance with AIX:iostat.IBM Technology Expert Labs/USA Copyright IBM Corporat
100、ion 202453SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the Responsiveness of Throughput SAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bre
101、ad+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objective:Demonstrating how%time_active affects the lon
102、gitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of systems storage at saturationIBM Technology Expert Labs/USA Copyright IBM Corporation 202454SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatOb
103、jective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the perform
104、ance of systems storage at saturationSAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/
105、USA Copyright IBM Corporation 202455SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitud
106、inalResponsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of systems storage at saturationSAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the trans
107、fers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202456SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O perfo
108、rmance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.Objective:
109、Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of
110、systems storage at saturationIBM Technology Expert Labs/USA Copyright IBM Corporation 202457SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_ac
111、tive of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202458SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8v
112、CPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the la
113、titudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202459SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsivene
114、ss%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202460SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 e
115、nt=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affe
116、cts the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202461SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Re
117、sponsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202462SPLPAR ventura:Power10 E1080 AIX
118、7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessAvg ServQ size%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_a
119、ctive affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202463SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ si
120、zeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202464SPLPAR ventura:Power10
121、E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating
122、how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202465SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAv
123、g ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202466SPLPAR ventur
124、a:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demo
125、nstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202467SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead Respon
126、sivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202468SPL
127、PAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Obje
128、ctive:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202469SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputR
129、ead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation
130、 202470SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the
131、drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202471SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O T
132、hroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM C
133、orporation 202472SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilizatio
134、n for the drive)Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objective:Demonstr
135、ating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of systems storage at saturationIBM Technology Expert Labs/USA Copyright IBM Corporation 202473SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating
136、SAN I/O performance with AIX:iostatObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughpu
137、t.Objective:Demonstrating the performance of systems storage at saturationSAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of b
138、read+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202474SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstratin
139、g how%time_active affects the latitudinal Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of systems storage at saturationSAN I/O performance=the Responsivenessof ThroughputSAN I/O perf
140、ormance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202475SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_thr
141、oughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the longitudinal
142、Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinal Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Obj
143、ective:Demonstrating the performance of systems storage at saturationIBM Technology Expert Labs/USA Copyright IBM Corporation 202476SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAv
144、g ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202477SPLPAR ventu
145、ra:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessAvg ServQ sizeWrite Responsiveness%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Dem
146、onstrating how%time_active affects the longitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202478SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead Resp
147、onsivenessWrite ResponsivenessAvg ServQ sizeSAN I/O performance=the service times of(rps+wps)of bread+bwrtn%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.
148、IBM Technology Expert Labs/USA Copyright IBM Corporation 202479SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessAvg ServQ sizeSAN I/O performance=the service times
149、 of(rps+wps)of bread+bwrtn%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202480SPLPAR ventura:Power
150、10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessAvg ServQ sizeSAN I/O performance=the service times of(rps+wps)of bread+bwrtn%time_active of the physical disk/tape over the 60sec int
151、erval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202481SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O perf
152、ormance with AIX:iostatSAN I/O performance=the service times of(rps+wps)of bread+bwrtnphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessAvg ServQ size%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active
153、 affects the longitudinal Responsiveness of Throughput.IBM Technology Expert Labs/USA Copyright IBM Corporation 202482SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the service times of(rps+wps)of bread+
154、bwrtnphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessAvg ServQ size%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.IBM Technology Expert Labs
155、/USA Copyright IBM Corporation 202483SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatSAN I/O performance=the service times of(rps+wps)of bread+bwrtnphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessAvg ServQ size
156、%time_active of the physical disk/tape over the 60sec interval(bandwidth utilization for the drive)Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating ho
157、w%time_active affects the latitudinal Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of systems storage at saturationIBM Technology Expert Labs/USA Copyright IBM Corporation 202484SPLP
158、AR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatObjective:Demonstrating how%time_active affects the Responsiveness of Throughput.Objective:Demonstrating how%time_active affects the latitudinalResponsiveness of Throughput.Objecti
159、ve:Demonstrating how%time_active affects the longitudinal Responsiveness of Throughput.Objective:Demonstrating the performance of systems storage at saturationSAN I/O performance=the Responsivenessof ThroughputSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferr
160、ed per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtnIBM Technology Expert Labs/USA Copyright IBM Corporation 202485PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Res
161、ponsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinalResponsiveness of ThroughputHow%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage
162、systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 202486PowerVM AIX:Monitoring SAN I/O to/from me
163、moryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinal Responsiveness of Thr
164、oughput How%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputI
165、BM Technology Expert Labs/USA Copyright IBM Corporation 202487SPLPAR wilshire:POWER9 E924 AIX 7.2 SMT8 ent=7.00 12vCPU vpm_throughput_mode=0Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullIBM Technology Expert Labs/US
166、A Copyright IBM Corporation 202488SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullIBM Technology Expert Labs/USA Copyright IBM Corporation
167、 202489SPLPAR figueroa:Power10 S1022 AIX 7.3 SMT8 16CPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullIBM Technology Expert Labs/USA Copyright IBM Corporation 202490SPLPAR fullerton:Power10 E108
168、0 AIX 7.1 SMT8 eCPU=20.00 20vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullIBM Technology Expert Labs/USA Copyright IBM Corporation 202491SPLPAR fullerton:Power10 E1080 AIX 7.1 SMT8 eCPU=20.
169、00 20vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputService Queue when fullservice queue full=1.0across multiple 60sec intervals begins degrading Responsivenessservice queue full=10.0across multiple 60sec intervals begins degrading Throughputservice qu
170、eue full=100.0 across multiple 60sec intervals is a critical performance issueIBM Technology Expert Labs/USA Copyright IBM Corporation 202492SPLPAR fullerton:Power10 E1080 AIX 7.1 SMT8 eCPU=20.00 20vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputService
171、 Queue when fullmilliseconds of latency only when the service queue is Technology Expert Labs/USA Copyright IBM Corporation 202493SPLPAR fullerton:Power10 E1080 AIX 7.1 SMT8 eCPU=20.00 20vCPU vpm_throughput_mode=2Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead Responsivenes
172、sWrite ResponsivenessService Queue when fullIBM Technology Expert Labs/USA Copyright IBM Corporation 202494SPLPAR washington:POWER9 E950 AIX 7.1 SMT8 eCPU=4.00 8vCPU vpm_throughput_mode=0Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessServi
173、ce Queue when fullIBM Technology Expert Labs/USA Copyright IBM Corporation 202495SPLPAR soledad:POWER9 E950 AIX 7.1 SMT8 eCPU=1.00 2vCPU vpm_throughput_mode=0Evaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullIBM Technol
174、ogy Expert Labs/USA Copyright IBM Corporation 202496Dedicated CPU LPAR alameda:Power10 E1080 AIX 7.3 SMT8 16CPUcores on Spectrum ScaleEvaluating SAN I/O performance with AIX:iostatphysical I/O ThroughputRead ResponsivenessWrite ResponsivenessService Queue when fullIBM Technology Expert Labs/USA Copy
175、right IBM Corporation 202497PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at sat
176、uration How%time_active affects the latitudinal Responsiveness of Throughput How%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from mem
177、oryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 202498PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatEx
178、plaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinal Responsiveness of Throughput How%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison an
179、d baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 202499SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throu
180、ghput_mode=2physical I/O ThroughputRead ResponsivenessWrite ResponsivenessSAN I/O performance=the service times of the transfers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtn+iostat-DRTl 60 10Service Queue when full
181、PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024100SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2physical I/O ThroughputRead ResponsivenessWrite ResponsivenessSAN I/O performance=the service times of the transf
182、ers per sec=(rps+wps)of bytes transferred per second(bps)=(bread+bwrtn)SAN I/O performance=the service times of(rps+wps)of bread+bwrtn+iostat-DRTl 60 10Service Queue when fullPowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024101SPLPAR ventura:P
183、ower10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+iostat-DRTl 60 10PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024102SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+iostat-DRTl 60 10PowerVM AIX:Mon
184、itoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024103SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024104SPLPAR ventura:Powe
185、r10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10IBM Technology Expert Labs/USA Copyright IBM Corporation 2024105SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80Po
186、werVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10+iostat-DRTl 60 10IBM Technology Expert Labs/USA Copyright IBM Corporation 2024106SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 1
187、0+iostat-DRTl 60 10IBM Technology Expert Labs/USA Copyright IBM Corporation 2024107SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10https:/ ASMIBM Technology Expert Labs/USA Copyright IBM Corporation 2024108S
188、PLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10https:/ Technology Expert Labs/USA Copyright IBM Corporation 2024109SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1
189、 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024110SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iost
190、at-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024111SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corpor
191、ation 2024112SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024113SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU v
192、pm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOhttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024114SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10A
193、SM|CIOhttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024115SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOhttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024116SPLPAR ven
194、tura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOhttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024117SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Moni
195、toring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOhttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024118SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOhttps:/ Technology Expert Labs
196、/USA Copyright IBM Corporation 2024119SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024120SPLPAR ventura:Power10 E1080 AIX
197、7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024121SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM A
198、IX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024122SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memoryASM|CIOIBM Technology Expert Labs/U
199、SA Copyright IBM Corporation 2024123SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memoryASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024124SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8v
200、CPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024125SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/
201、O to/from memory+uptime;ps guwASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024126SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memoryASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corpora
202、tion 2024127SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+ps kf|sort nk4IBM Technology Expert Labs/USA Copyright IBM Corporation 2024128SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM
203、AIX:Monitoring SAN I/O to/from memoryThe nonzero values of AIX:ps-kf|sort-nk4:C account the CPU clock ticks of onlythe executing sys PID:TIDs over the last Sec;C is not cumulative between Secs.As such,the nonzero C values define the real-time sys workload of the last Sec.IBM Technology Expert Labs/U
204、SA Copyright IBM Corporation 2024129SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024130SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=
205、2PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024131SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memory+ps klmo THREADIBM Technology Expert Labs/USA Copyright IBM Corpora
206、tion 2024132SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryThe nonzero values of AIX:ps-kf|sort-nk4:C account the CPU clock ticks of onlythe executing sys PID:TIDs over the last Sec;C is not cumulative between Secs.As such,t
207、he nonzero C values define the real-time sys workload of the last Sec.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024133SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IB
208、M Corporation 2024134SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024135SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Mo
209、nitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024136SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memoryASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2
210、024137SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memoryASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024138SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2Power
211、VM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024139SPLPAR ventura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryIBM Technology Expert Labs/USA Copyright IBM Corporation 2024140SPLPAR ve
212、ntura:Power10 E1080 AIX 7.2 SMT8 ent=8.00 8vCPU vpm_throughput_mode=2+vmstat-IWwt 1 80PowerVM AIX:Monitoring SAN I/O to/from memory+iostat-DRTl 60 10ASM|CIOIBM Technology Expert Labs/USA Copyright IBM Corporation 2024141PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/
213、O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturation How%time_active affects the latitudinal Responsiveness of Throughput How%time_active affects the lon
214、gitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright
215、 IBM Corporation 2024142PowerVM AIX:Monitoring SAN I/O to/from memoryMonitoring SAN I/OEvaluating SAN I/O performance with AIX:iostatDistinguishing Throughput and Responsiveness in AIX:iostatExplaining sqfull(service queue full)in AIX:iostatDemonstrating the performance of systems storage at saturat
216、ion How%time_active affects the latitudinal Responsiveness of Throughput How%time_active affects the longitudinal Responsiveness of ThroughputA parade of other storage systems for comparison and baselineMonitoring SAN I/O to/from memoryThe performance of systems storage at saturation to/from memoryA
217、 common cause of systems storage under-utilization of ThroughputIBM Technology Expert Labs/USA Copyright IBM Corporation 2024143PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:A
218、ll SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput a
219、nd responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage technologies,i.e.,IBM FlashSystemsthe immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.All SAN Reads must be received in real memory and all SAN Writes must b
220、e staged in real memory before flushing to SAN but how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the responsiveness of 1.0ms average service times,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.IBM Technology Expert Lab
221、s/USA Copyright IBM Corporation 2024144SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryGiven the responsiveness of 1.0ms average service times,the immediate readiness of memory to support SAN I/O is an unseen emerging factor
222、 in performance.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024145SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryGiven the responsiveness of 1.0ms average service times,the immediate readiness of memory to sup
223、port SAN I/O is an unseen emerging factor in performance.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024146SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryGiven the responsiveness of 1.0ms average service times
224、,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024147SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaultJF
225、S2 default I/OIBM Technology Expert Labs/USA Copyright IBM Corporation 2024148SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaultJFS2 default I/OIBM Technology Expert Labs/USA Copyright IBM Corporation 2024149SPLPAR
226、 roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaulthttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024150SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring
227、SAN I/O to/from memoryJFS2 defaulthttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024151SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaulthttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2
228、024152SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaulthttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024153SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX
229、:Monitoring SAN I/O to/from memoryhttps:/ Technology Expert Labs/USA Copyright IBM Corporation 2024154SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryhttps:/ is the subset of w that are blocked IOs on the VMM wait queueIBM T
230、echnology Expert Labs/USA Copyright IBM Corporation 2024155SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryhttps:/ is the subset of w that are blocked IOs on the VMM wait queueIBM Technology Expert Labs/USA Copyright IBM Cor
231、poration 2024156SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaulthttps:/ is the subset of w that are blocked IOs on the VMM wait queue+IBM Technology Expert Labs/USA Copyright IBM Corporation 2024157SPLPAR roscoe:
232、Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaulthttps:/ is the subset of w that are blocked IOs on the VMM wait queue+IBM Technology Expert Labs/USA Copyright IBM Corporation 2024158SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent
233、=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaulthttps:/ is the subset of w that are blocked IOs on the VMM wait queue+IBM Technology Expert Labs/USA Copyright IBM Corporation 2024159SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_m
234、ode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaultIBM Technology Expert Labs/USA Copyright IBM Corporation 2024160SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default?IBM Technology Expert Labs/USA Copyri
235、ght IBM Corporation 2024161SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default?IBM Technology Expert Labs/USA Copyright IBM Corporation 2024162SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput
236、_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 defaultIBM Technology Expert Labs/USA Copyright IBM Corporation 2024163SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default|rbr|rbw I/OIBM Technology Expert La
237、bs/USA Copyright IBM Corporation 2024164SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default|rbr|rbw I/OIBM Technology Expert Labs/USA Copyright IBM Corporation 2024165SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.
238、00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default I/OAll SAN Reads must be received in real memory and all SAN Writes must be staged in real memorybefore flushing to SAN.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024166SPLPAR roscoe:Power10 E1050
239、 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default I/OAll SAN Reads must be received in real memory and all SAN Writes must be staged in real memorybefore flushing to SAN but how quickly memory receives SAN Reads and stages SAN writes is ign
240、ored.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024167SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryGiven the responsiveness of 1.0ms average service times,the immediately available memory to support SAN I/O
241、 is fre(free memory).JFS2 default I/OIBM Technology Expert Labs/USA Copyright IBM Corporation 2024168SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default I/OAIX:lrudGiven the responsiveness of 1.0ms average service t
242、imes,the strenuously slower memory to support SAN I/O is AIX:lrud:fr(free rate memory).IBM Technology Expert Labs/USA Copyright IBM Corporation 2024169SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default I/OAIX:lrudG
243、iven the responsiveness of 1.0ms average service times,the strenuously slower memory to support SAN I/O is AIX:lrud:sr(scan rate overhead).IBM Technology Expert Labs/USA Copyright IBM Corporation 2024170SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monito
244、ring SAN I/O to/from memoryJFS2 default I/OAIX:lrudhttps:/ the responsiveness of 1.0ms average service times,the strenuously slower memory to support SAN I/O is AIX:lrud(fr sr page-replacement algorithm).IBM Technology Expert Labs/USA Copyright IBM Corporation 2024171SPLPAR roscoe:Power10 E1050 AIX
245、7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default I/OAIX:lrudGiven the responsiveness of 1.0ms average service the strenuously slower memory to support SAN I/O is AIX:lrud(fr sr page-replacement algorithm).IBM Technology Expert Labs/USA Copyrigh
246、t IBM Corporation 2024172SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryJFS2 default I/OAIX:lrudGiven the responsiveness of 1.0ms average service the strenuously slower memory to support SAN I/O is AIX:lrud(fr sr page-repla
247、cement algorithm).SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT8 ent=18.00 18vCPU vpm_throughput_mode=2IBM Technology Expert Labs/USA Copyright IBM Corporation 2024173SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudGiven t
248、he responsiveness of 1.0ms average service times,the immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.JFS2 default I/OIBM Technology Expert Labs/USA Copyright IBM Corporation 2024174SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_m
249、ode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudJFS2 default I/OIBM Technology Expert Labs/USA Copyright IBM Corporation 2024175SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudJFS2 default I/O?AIX:lrud?IBM
250、Technology Expert Labs/USA Copyright IBM Corporation 2024176SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrud?JFS2 default I/O?AIX:lrud?IBM Technology Expert Labs/USA Copyright IBM Corporation 2024177SPLPAR roscoe:Pow
251、er10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudb is the subset of w that are blocked IOs on the VMM wait queue!IBM Technology Expert Labs/USA Copyright IBM Corporation 2024178SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU
252、 vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudb is the subset of w that are blocked IOs on the VMM wait queue!AIX:lrudIBM Technology Expert Labs/USA Copyright IBM Corporation 2024179SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AI
253、X:Monitoring SAN I/O to/from memoryAIX:lrudJFS2 default I/OJFS2 defaultAll SAN Reads must be received in real memory and all SAN Writes must be staged in real memorybefore flushing to SAN but how quickly memory receives SAN Reads and stages SAN writes is ignored.IBM Technology Expert Labs/USA Copyri
254、ght IBM Corporation 2024180SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudJFS2 default I/OJFS2 defaultGiven the responsiveness of 1.0ms average service times,the immediate readiness of memory to support SAN I/O is a
255、n unseen emerging factor in performance.IBM Technology Expert Labs/USA Copyright IBM Corporation 2024181SPLPAR roscoe:Power10 E1050 AIX 7.2 SMT4 ent=18.00 36vCPU vpm_throughput_mode=2PowerVM AIX:Monitoring SAN I/O to/from memoryAIX:lrudJFS2 default I/OJFS2 defaultIBM Technology Expert Labs/USA Copyr
256、ight IBM Corporation 2024182PowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before
257、 flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 1.0ms average service times in todays OEM systems storage te
258、chnologies,i.e.,IBM FlashSystemsthe immediate readiness of memory to support SAN I/O is an unseen emerging factor in performance.All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN but how quickly memory receives SAN Reads and stages
259、SAN writes is ignored.Given the responsiveness of 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers of the PowerVM CPU nomenclature at SMT8PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Minimizing CPUtime recycling at SMT8Re
260、ducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memoryEvaluating SAN I/Oa999011 PowerVM AIX:Mapping the path of workload executionDeep Science monitoringPowerVM AIX:Monitoring SAN I/O to/from memoryABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice
261、compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushing to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is
262、 ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers of the PowerVM CPU nomenclature at SMT8 PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Mi
263、nimizing CPUtime recycling at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memory Evaluating SAN I/Oa999011 PowerVM AIX:Mapping the path of workload execution Deep Science monitoringABSTRACTThis session begins a 6-part curriculum about a problematic failing at SMT8:Runni
264、ng beyond CPU entitlement when CPU is 50%idle.How can AIX exhaust CPU entitlement when CPU is 50%idle?Most of us monitor AIX-only(i.e.,AIX:nmon)with no awareness of PowerVM.This curriculum is about monitoring the deeper numbers of AIX-on-PowerVM(i.e.,AIX:lparstat-H:cede)and with this ability,we can
265、adapt our practices to better exploit Power910 AIX at SMT8 as intended per design.Theme of the 2024 curriculum:Processing tighter on fewer software licensesPowerVM AIX:Running beyond CPU entitlement when CPU is 50%idleProblematic ThemeIBM Technology Expert Labs/USA Copyright IBM Corporation 2024209a
266、999006 PowerVM AIX:Running beyond CPU entitlement when CPU is 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers of the PowerVM CPU nomenclature at SMT8 PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Minimizing CPUtime recycl
267、ing at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memory Evaluating SAN I/Oa999011 PowerVM AIX:Mapping the path of workload execution Deep Science monitoringABSTRACTThis session begins with a conceptual explanation of each layer of the PowerVM CPU nomenclaturebefore in
268、troducing the realization of CPUtime recycling at SMT8.This session is deconstructed from what AIX shows us about AIX-on-PowerVM;this session is based on what AIX shows us about each layer supporting AIX-on-PowerVM.Understanding this structure is the first step towards implementing the tactical prac
269、tices detailed in the next two sessions.Beyond what AIX shows us,there are no references about:The 6 layers of the PowerVM CPU nomenclature at SMT8.As such,this session is unprecedented and perhaps shockingly revealing.PowerVM AIX:The 6 layers of the PowerVM CPU nomenclature at SMT8IBM Technology Ex
270、pert Labs/USA Copyright IBM Corporation 2024210a999006 PowerVM AIX:Running beyond CPU entitlement when CPU is 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers of the PowerVM CPU nomenclature at SMT8PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUti
271、mea999009 PowerVM AIX:Minimizing CPUtime recycling at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memory Evaluating SAN I/Oa999011 PowerVM AIX:Mapping the path of workload execution Deep Science monitoringPowerVM AIX:Monitoring CPUtime recycling at SMT8ABSTRACTThis sess
272、ion will detail the tactics for:Monitoring CPUtime recycling at SMT8Most of us monitor AIX-only with no awareness of PowerVM,i.e.AIX:nmon.This session is about monitoring AIX-on-PowerVM with AIX:lparstat-h|-H and AIX:mpstat-v.This session will distinguish Used vs Accessible vs Available CPUtime in A
273、IX-on-PowerVM and then detail how Accessible CPUtime is excess CPUtime in AIX:vmstat:(cpu:pc).IBM Technology Expert Labs/USA Copyright IBM Corporation 2024211a999006 PowerVM AIX:Running beyond CPU entitlement when CPU is 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers of the PowerVM CPU no
274、menclature at SMT8 PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Minimizing CPUtime recycling at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memory Evaluating SAN I/Oa999011 PowerVM AIX:Mapping the pat
275、h of workload executionDeep Science monitoringABSTRACTThis session will detail the tactics for:Minimizing CPUtime recycling at SMT8Excess CPUtime is recycled in AIX:vmstat:(cpu:pc).When excess CPUtime is sustained at greater than 50%of AIX:vmstat:(cpu:pc)then:AIX:vmstat:(cpu:pc)is over-inflated with
276、 too much recycled excess CPUtime;there are too many configured virtual CPUs(vCPUs)with too little productive workload;too many software licenses are processing loosely on too many virtual CPUs(vCPUs)Minimizing CPUtime recycling at SMT8 leads to Processing tighter on fewer software licenses.Theme of
277、 the 2024 curriculum:Processing tighter on fewer software licenses PowerVM AIX:Minimizing CPUtime recycling at SMT8IBM Technology Expert Labs/USA Copyright IBM Corporation 2024212a999006 PowerVM AIX:Running beyond CPU entitlement when CPU is 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers
278、of the PowerVM CPU nomenclature at SMT8 PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Minimizing CPUtime recycling at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memoryEvaluating SAN I/Oa999011 PowerVM
279、 AIX:Mapping the path of workload executionDeep Science monitoringABSTRACTEvaluating SAN I/O performance with AIX:iostat is an easy direct practice compared to the other end.Most of us forget:All SAN Reads must be received in real memory and all SAN Writes must be staged in real memory before flushi
280、ng to SAN,i.e.,real memory is the other end.Compared to SAN I/O monitoring,evaluating how quickly memory receives SAN Reads and stages SAN writes is ignored.Given the extraordinary throughput and responsiveness of SSD/Flash media at 50%idle Problematic Themea999007 PowerVM AIX:The 6 layers of the Po
281、werVM CPU nomenclature at SMT8 PowerVM Conceptsa999008 PowerVM AIX:Monitoring CPUtime recycling at SMT8Defining excess CPUtimea999009 PowerVM AIX:Minimizing CPUtime recycling at SMT8Reducing excess CPUtimea999010 PowerVM AIX:Monitoring SAN I/O to/from memory Evaluating SAN I/Oa999011 PowerVM AIX:Map
282、ping the path of workload executionDeep Science monitoringABSTRACTThis session is about Deep Science monitoring.This session is not about monitoring with no sense of PowerVM(i.e.,AIX:nmon);rather,this session unveils the subtleties of AIX-on-PowerVM.This session traces how the 6 layers of the PowerV
283、M CPU nomenclature at SMT8 was deconstructed from what AIX shows us about AIX-on-PowerVM.This session then uses AIX command output to map the path of workload execution through the 6 layers of the PowerVM CPU nomenclature at SMT8.Beyond what AIX shows us,there are no references about AIX-on-PowerVM.As such,this session is unprecedented and perhaps shockingly revealing.PowerVM AIX:Mapping the path of workload execution
0731-84720580
sgpjbg002
工作日 9:30 - 18:00
研報
新質生產力
DeepSeek?
低空經濟
大模型
AI Agent
AI Infra
具身智能
自動駕駛
寵物
銀發經濟
人形機器人
企業出海
算力
微短劇
薪酬
白皮書
創新藥
人工智能
AI產業
信息科技
互聯網
消費經濟
汽車交通
電商零售
傳媒娛樂
醫藥健康
投資金融
能源環境
地產建筑
傳統產業
英文報告
其它
最新報告
中英對照
全文搜索
報告精選
PDF上傳翻譯
多格式文檔互轉
入駐&報告售賣
會員權益
機構報告
券商研報
財報庫
專題合集
英文報告
數據圖表
會議報告
其他資源
行業聚焦
芯片產業
熱點概念
全球咨詢智庫
人工智能
500強
新質生產力
會議峰會
新能源汽車
企業年報
互聯網
公司研究
行業綜觀
消費教育
科技通信
醫藥健康
人力資源
投資金融
汽車產業
物流地產
電子商務
傳統產業
傳媒營銷
其它
2025低空經濟/低空產業報告合集（共49套打包）
2025年智算中心/AIDC/AI數據中心研究報告合集（共37套打包）
AI、科技與通信
廣告、傳媒與營銷
消費、零售與支付
HR、文化與旅游
金融、保險與投資
能源、環境與工業
醫療制藥與大健康
物流、地產與建筑
其他行業
AI ? 科技 ? 通信
數字化
金融財經
智能制造
電商傳媒
地產建筑
醫療醫學
能源化工
其他行業
收藏
下載
2025-05-20
AI查數
行業數據
政策法規
商業模式
產業鏈
競爭格局
市場規模
產業概述
其它
2025年
AI讀財報
年報
一季報
半年報
三季報
IPO招股書
社會責任報告
A股
IPO申報
港股
美股&全球
新三板
掃碼登錄
手機快捷登錄
賬號登錄
