Spindles / SAN

[Dave_O]

*Creating a VMFS datastore

*Expanding a VMFS datastore onto additional extents

*Powering on a virtual machine

*Acquiring a lock on a file

*Creating or deleting a file

*Creating a template

*Deploying a virtual machine from a template

*Creating a new virtual machine

Migrating a virtual machine with VMotion (host or storage?)

Growing a file, for example, a Snapshot file or a thin provisioned Virtual Disk

Any software that updates itself regularly, particularly system utilities like antivirus programs and operating system updates

 

Looking at these the ones starred don't happen very often nor do they take very long. So what are the advantages?

[oxide54]

Host VMOTION would benefit from reducing LUN locking, Storage VMotion would benefit from the SAN copying as well.

 

Snapshots lock the LUN every 16mb of growth. not sure on thin provsioned .

 

i think the last item they listed is related to thin provisoned disks.

 

for me the SAN to SAN copy is the reason. also reduces CPU load on the ESX box as well.

 

its also not so much about speeding up provisoning a VM for the administrator, but not degrading the users performance as a result.

[Dave_O]

SAN to SAN copying I do once every 5 years (when I decomission a unit), funnily enough just last month! I know it can mean LUN to LUN on the same box. I'll have a look at it, I know the V7000 supports it.probably have a mess around with it at the weekend

[Dave_O]

Well I'll be! Just checked VAAI is on, supported and working and I didn't know! The wonders of modern technology eh! I'll give you a thank you for that one (even though technically nothing has changed - just makes me feel better knowing more) Edited February 20, 2012 by Dave_O

[oxide54]

maybe why your not seeing any performance problems with lun locking???

[Dave_O]

I installed SQL IO on my SIMS server (virtual) and ran 4 tests - sequential read & write and random read & write over a 30 second period. The results are below. I would be interested in the results of others running the same commands (also given below):-

 

sqlio -kR -s30 -frandom -o8 -b8 -LS -Fparam.txt

CUMULATIVE DATA:

throughput metrics:

IOs/sec: 34867.99

MBs/sec: 272.40

latency metrics:

Min_Latency(ms): 0

Avg_Latency(ms): 0

Max_Latency(ms): 98

histogram:

ms: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+

%: 99 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

sqlio -kW -s30 -frandom -o8 -b8 -LS -Fparam.txt

CUMULATIVE DATA:

throughput metrics:

IOs/sec: 34123.80

MBs/sec: 266.59

latency metrics:

Min_Latency(ms): 0

Avg_Latency(ms): 0

Max_Latency(ms): 84

histogram:

ms: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+

%: 99 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

sqlio -kR -s30 -fsequential -o8 -b8 -LS -Fparam.txt

CUMULATIVE DATA:

throughput metrics:

IOs/sec: 36841.62

MBs/sec: 287.82

latency metrics:

Min_Latency(ms): 0

Avg_Latency(ms): 0

Max_Latency(ms): 23

histogram:

ms: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+

%: 99 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

sqlio -kW -s30 -fsequential -o8 -b8 -LS -Fparam.txt

CUMULATIVE DATA:

throughput metrics:

IOs/sec: 10728.56

MBs/sec: 83.81

latency metrics:

Min_Latency(ms): 0

Avg_Latency(ms): 0

Max_Latency(ms): 52

histogram:

ms: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24+

%: 85 0 0 0 2 3 2 4 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

 

It's just over a year since I posted this. Has anyone else done this test? I would be interested in seeing the results both from a physical and virtual SIMS instance