Introduction:
In current hybrid cloud era, majority of the customers are looking for their infrastructure to be in public cloud and they would like to invest more on workloads rather than maintaining infrastructure. When we have workloads running in hybrid cloud, data becomes crucial, and it also depends on how much IOPS speed workload is needed. Also, customers would like to move data from one IOPS speed to other IOPS speed based on their workload characteristics. For example, customer would like to run SAP HANA workload would like to run with higher IOPS, meanwhile same customer would like to run devOPs related workload on lower IOPS.
Need of flexible IOPS feature :
Power Virtual Server (PowerVS) offers variety of hybrid cloud features in hybrid cloud environment. In variety of features, live data migration from one IOPS speed to other IOPS speed is released in 4Q-2024. With this feature, customers can move their workload from one IOPS to other IOPS. For example, if customer has deployed their VMs devOPs workload in high IOPS and this workload doesn’t need high IOPS. With this feature, they can move their workload from high IOPS to low IOPS and this is live data migration. With this, customer can make sure their billing is less for low IOPS workload.
This blog explains how customers can use Flexible IOPS feature and how we can seamlessly move data from one IOPS speed to other IOPS without hampering any impact to workload.
Flexible IOPS feature with different tiers:
PowerVS has different tiers for different IOPS speed and there is different billing for different tiers.
- · tier0 is with speed 25 IOPS
- · tier1 is with speed 10 IOPS
- · tier3 is with speed 3 IOPS
- · tier5k (AKA Flexible IOPS) is with speed 5000 IOPS
With Flexible IOPS feature, one can move volume from one tier to other tier and there by the data present in those volumes will be moved.
Demo on moving volume from one tier to other tier with workload running
In this blog, we are running IO workload on VM and showcasing how we can move volume attached to VM from one tier to other tier.
- Deployed an AIX VM from IBM Cloud UI
As we can from above snapshot , we are deploying VM with boot volume in tier3 and data volumes in tier5k. Now, VM is in a position to run volumes spread across 2 different tiers.
-
Attaching volumes from all 4 different tiers
With above snapshot, we can see we are attaching tier0 volume to VM which we deployed in step 1.
- Final view of VM which has data volumes attached from all 4 different tiers.
- We have started IO workload inside VM:
[root@dnd-srik] /
# hostname;what /unix|grep build;oslevel -s
dnd-srik
_kdb_buildinfo unix_64 Apr 21 2023 13:58:32 2316B_73B
7300-01-02-2320
[root@dnd-srik] /
#
[root@dnd-srik] /
# lspv
hdisk0 none None
hdisk1 00fa00d66c59c9d7 rootvg active
hdisk2 none None
hdisk3 none None
hdisk4 none None
hdisk5 none None
hdisk6 none None
hdisk7 none None
hdisk8 none None
hdisk9 none None
hdisk10 none None
hdisk11 none None
hdisk12 none None
hdisk13 none None
hdisk14 none None
hdisk15 none None
hdisk16 none None
hdisk17 none None
hdisk18 none None
hdisk19 none None
hdisk20 none None
hdisk21 none None
[root@dnd-srik] /
#
[root@dnd-srik] /tmp/paws_2.4_vios_ci
# iostat -d 2
System configuration: lcpu=8 drives=23 paths=176 vdisks=1
Disks: % tm_act Kbps tps Kb_read Kb_wrtn
hdisk19 100.0 400.0 100.0 512 288
hdisk15 100.0 400.0 100.0 472 328
hdisk21 100.0 398.0 99.5 480 316
hdisk16 100.0 400.0 100.0 488 312
hdisk17 100.0 398.0 99.5 460 336
hdisk14 100.0 398.0 99.5 512 284
hdisk20 100.0 400.0 100.0 456 344
hdisk18 100.0 398.0 99.5 480 316
hdisk13 100.0 398.0 99.5 492 304
hdisk12 100.0 398.0 99.5 472 324
hdisk11 100.0 400.0 99.5 528 272
cd0 0.0 0.0 0.0 0 0
hdisk7 2.0 340.0 85.0 396 284
hdisk8 1.5 428.0 107.0 536 320
hdisk10 1.0 352.0 88.0 396 308
hdisk9 1.5 394.0 98.5 408 380
hdisk6 2.0 370.0 92.5 456 284
hdisk1 0.0 0.0 0.0 0 0
hdisk2 0.5 386.0 96.5 456 316
hdisk4 1.5 402.0 100.5 468 336
hdisk0 0.0 0.0 0.0 0 0
hdisk3 3.0 354.0 88.5 396 312
hdisk5 2.0 384.0 96.0 464 304
Disks: % tm_act Kbps tps Kb_read Kb_wrtn
hdisk19 100.0 398.0 99.5 496 300
hdisk15 100.0 396.0 99.0 460 332
hdisk21 100.0 398.0 99.5 476 320
hdisk16 100.0 398.0 99.5 496 300
hdisk17 100.0 398.0 99.5 492 304
hdisk14 100.0 398.0 99.5 432 364
hdisk20 100.0 398.0 99.5 472 324
hdisk18 100.0 400.0 100.0 472 328
hdisk13 100.0 398.0 99.5 492 304
hdisk12 100.0 400.0 100.0 488 312
hdisk11 100.0 400.0 100.0 492 308
cd0 0.0 0.0 0.0 0 0
hdisk7 3.5 376.0 94.0 464 288
hdisk8 1.5 366.0 91.5 452 280
hdisk10 3.5 376.0 94.0 460 292
hdisk9 1.5 372.0 93.0 432 312
hdisk6 3.0 402.0 100.5 452 352
hdisk1 0.0 0.0 0.0 0 0
hdisk2 1.5 364.0 91.0 432 296
hdisk4 0.5 372.0 93.0 464 280
hdisk0 0.0 0.0 0.0 0 0
hdisk3 3.5 418.0 104.5 456 380
hdisk5 2.0 370.0 92.5 480 260
Topas Monitor for host:dnd-srik EVENTS/QUEUES FILE/TTY
Wed Jun 19 07:38:18 2024 Interval:2 Cswitch 8199 Readch 7330.3K
Syscall 53784 Writech 5641.0K
CPU User% Kern% Wait% Idle% Physc Entc% Reads 1835 Rawin 0
Total 4.6 9.9 18.8 66.7 0.17 66.35 Writes 803 Ttyout 743
Forks 0 Igets 0
Network BPS I-Pkts O-Pkts B-In B-Out Execs 0 Namei 2
Total 1008. 1.50 2.00 105.5 902.5 Runqueue 0 Dirblk 0
Waitqueue 0.0
Disk Busy% BPS TPS B-Read B-Writ MEMORY
hdisk10 1.5 460K 115.0 262K 198K PAGING Real,MB 3072
hdisk6 2.5 448K 112.0 282K 166K Faults 10 % Comp 61
hdisk13 4.5 430K 107.5 258K 172K Steals 0 % Noncomp 7
hdisk4 3.5 410K 102.5 236K 174K PgspIn 0 % Client 7
hdisk20 100.0 402K 100.0 224K 178K PgspOut 0
hdisk14 100.0 400K 100.0 256K 144K PageIn 0 PAGING SPACE
hdisk18 100.0 400K 100.0 242K 158K PageOut 0 Size,MB 512
hdisk8 2.5 400K 100.0 234K 166K Sios 0 % Used 7
hdisk12 100.0 400K 100.0 252K 148K % Free 93
hdisk19 100.0 400K 99.50 252K 148K NFS (calls/sec)
hdisk11 100.0 400K 100.0 218K 182K SerV2 0 WPAR Activ 0
hdisk21 100.0 398K 99.50 224K 174K CliV2 0 WPAR Total 0
hdisk15 100.0 398K 99.50 248K 150K SerV3 0 Press: "h"-help
hdisk17 100.0 398K 99.50 242K 156K CliV3 0 "q"-quit
hdisk16 100.0 398K 99.50 244K 154K SerV4 0
hdisk5 2.0 394K 98.50 250K 144K CliV4 0
hdisk7 2.5 392K 98.00 248K 144K
hdisk2 3.0 390K 97.50 246K 144K
hdisk9 2.0 388K 97.00 242K 146K
hdisk3 1.5 382K 95.50 230K 152K
hdisk1 0.0 0 0 0 0
hdisk0 0.0 0 0 0 0
cd0 0.0 0 0 0 0
FileSystem BPS TPS B-Read B-Writ
Total 743.5 0.50 743.5 0
Name PID CPU% PgSp Owner
filemon 11993834 4.9 5.24M root
vfc_kpro 2163014 2.9 4.19M root
topas 43581762 0.7 13.9M root
trclogio 11862702 0.1 512K root
xmgc 852254 0.0 448K root
getty 11665770 0.0 668K root
randtime 66912520 0.0 1.34M root
randtime 59113754 0.0 1.76M root
randtime 56033724 0.0 1.94M root
gil 1573172 0.0 960K root
randtime 19923310 0.0 3.55M root
randtime 66781444 0.0 1.34M root
randtime 27918690 0.0 2.72M root
randtime 26476854 0.0 3.78M root
randtime 55706034 0.0 1.96M root
randtime 5571256 0.0 1016K root
randtime 47448502 0.0 4.59M root
randtime 18743632 0.0 2.82M root
randtime 57999864 0.0 1.83M root
- Now, with workload running on VM, we are moving volumes from one tier to other tier. For example, we are moving volume in tier1 to tier5k (Flexible IOPS):
- Another example, where we are moving another volume from tier5k to tier3:
- Now, lets check the workload on VM:
Topas Monitor for host:dnd-srik EVENTS/QUEUES FILE/TTY
Wed Jun 19 07:41:12 2024 Interval:2 Cswitch 8262 Readch 7228.1K
Syscall 53647 Writech 5518.7K
CPU User% Kern% Wait% Idle% Physc Entc% Reads 1809 Rawin 0
Total 4.5 9.7 6.5 79.2 0.16 65.06 Writes 772 Ttyout 633
Forks 0 Igets 0
Network BPS I-Pkts O-Pkts B-In B-Out Execs 0 Namei 3
Total 898.0 1.50 2.00 97.00 801.0 Runqueue 2.50 Dirblk 0
Waitqueue 0.0
Disk Busy% BPS TPS B-Read B-Writ MEMORY
hdisk8 3.0 446K 111.5 286K 160K PAGING Real,MB 3072
hdisk9 3.0 416K 104.0 250K 166K Faults 0 % Comp 61
hdisk16 100.0 400K 100.0 222K 178K Steals 0 % Noncomp 7
hdisk2 3.0 400K 100.0 248K 152K PgspIn 0 % Client 7
hdisk3 3.0 400K 100.0 226K 174K PgspOut 0
hdisk14 99.5 400K 100.0 252K 148K PageIn 0 PAGING SPACE
hdisk13 100.0 400K 100.0 244K 156K PageOut 0 Size,MB 512
hdisk20 100.0 400K 100.0 244K 156K Sios 0 % Used 7
hdisk12 100.0 398K 99.50 254K 144K % Free 93
hdisk19 100.0 398K 99.50 252K 146K NFS (calls/sec)
hdisk11 100.0 398K 99.50 230K 168K SerV2 0 WPAR Activ 0
hdisk15 100.0 398K 99.50 236K 162K CliV2 0 WPAR Total 0
hdisk18 100.0 396K 99.00 254K 142K SerV3 0 Press: "h"-help
hdisk10 2.5 394K 98.50 224K 170K CliV3 0 "q"-quit
hdisk6 0.5 380K 95.00 248K 132K SerV4 0
hdisk5 3.5 378K 94.50 234K 144K CliV4 0
hdisk4 3.0 374K 93.50 226K 148K
hdisk21 2.0 374K 93.50 228K 146K
hdisk7 2.0 358K 89.50 218K 140K
hdisk17 2.0 354K 88.50 210K 144K
hdisk1 0.0 0 0 0 0
hdisk0 0.0 0 0 0 0
cd0 0.0 0 0 0 0
FileSystem BPS TPS B-Read B-Writ
Total 633.5 0.50 633.5 0
Name PID CPU% PgSp Owner
filemon 11993834 5.4 5.27M root
vfc_kpro 2163014 2.8 4.19M root
topas 43581764 0.7 13.9M root
trclogio 11862702 0.0 512K root
getty 11665770 0.0 668K root
randtime 27001158 0.0 3.79M root
randtime 62587268 0.0 1.57M root
randtime 2032204 0.0 1.21M root
randtime 28442994 0.0 3.86M root
randtime 12059132 0.0 3.24M root
randtime 21496222 0.0 2.58M root
randtime 59900210 0.0 1.72M root
randtime 47382964 0.0 4.59M root
randtime 64160180 0.0 1.49M root
randtime 36766064 0.0 4.18M root
randtime 63570338 0.0 1.52M root
randtime 24642046 0.0 3.19M root
randtime 55116192 0.0 1.99M root
randtime 17826096 0.0 3.48M root
randtime 12321234 0.0 3.25M root
From above, we can see workload is continuously running even though we moved volume from one tier to other tier
Volume movement from one tier to other tier via IBM Cloud CLI :
Using ibmcloud pi vol action command, we can move volume from one tier to other tier.
For example :
srikanthjoshi@Srikanths-MacBook-Pro ~ % ibmcloud pi vol get edab821e-f4b8-4ee5-903c-78432b7a8520
Getting Volume edab821e-f4b8-4ee5-903c-78432b7a8520 under account PPCaaS Test Account as user srikanth.joshi@in.ibm.com...
ID edab821e-f4b8-4ee5-903c-78432b7a8520
Name DND-srik-vol-tier1
Profile tier5k
Status in-use
Size 1
Created 2024-06-19T09:28:49.000Z
Updated 2024-06-19T12:40:06.000Z
Shareable false
Bootable false
Storage Pool general-flash-2
IO Throttle Rate 5000 iops
PVMInstanceIDs 7eacf3b1-7d0f-4038-ad12-a0393617eeef
WWN 600507681284817BE80000000001EDDA
srikanthjoshi@Srikanths-MacBook-Pro ~ % ibmcloud pi vol action edab821e-f4b8-4ee5-903c-78432b7a8520 --target-tier tier0
Performing action on volume edab821e-f4b8-4ee5-903c-78432b7a8520 under account PPCaaS Test Account as user srikanth.joshi@in.ibm.com...
OK
Action on Volume ID edab821e-f4b8-4ee5-903c-78432b7a8520 successful.
srikanthjoshi@Srikanths-MacBook-Pro ~ % ibmcloud pi vol get edab821e-f4b8-4ee5-903c-78432b7a8520
Getting Volume edab821e-f4b8-4ee5-903c-78432b7a8520 under account PPCaaS Test Account as user srikanth.joshi@in.ibm.com...
ID edab821e-f4b8-4ee5-903c-78432b7a8520
Name DND-srik-vol-tier1
Profile tier0
Status in-use
Size 1
Created 2024-06-19T09:28:49.000Z
Updated 2024-06-19T12:46:52.000Z
Shareable false
Bootable false
Storage Pool general-flash-2
IO Throttle Rate 25 iops
PVMInstanceIDs 7eacf3b1-7d0f-4038-ad12-a0393617eeef
WWN 600507681284817BE80000000001EDDA
srikanthjoshi@Srikanths-MacBook-Pro ~ %
Advantages:
As we can see, customers will have huge advantage of moving volumes from one IOPS to other IOPS with data present on them. This will help customers to reduce their billing for workload which they don’t need high IOPS and it also helps to make sure volume movement from one tier to other tier is independent of workload running on VM.
References:
· Power Virtual Server overview : https://www.ibm.com/products/power-virtual-server
· Power Virtual Server API Doc : https://cloud.ibm.com/apidocs/power-cloud