3PAR sent me a report saying my disks are exceeding their IOPS threshold.
I was given the following figures (IOPS per spindle)
Nearline - 75
FC 10k - 150
FC 15k - 200
If I do a performance report of Physical Disk Usage - Total IOPS, it looks like my Nearline is averaging between 175-200 IOPS per disk.
The Nearline disk is used in ESX for one File Server. This server processes video so it has heavy read writes all day.
So the question, does this about right? And is the only solution faster disk and/or more spindles?
TIA
IOPS per Physical Disk?
Re: IOPS per Physical Disk?
I've been doing some performance testing (simple file copy within Windows Explorer) and am having mixed results.
I have two VV's, one FC, one NL presented to the fileserver. While no other activity is happening on the array, I copy from one volume to the other and back again numerous times (only one copy job running at any one time).
The results are ranging from 20MB/s up to 200MB/s for a 1.5GB test file.
Can anyone explain why such a wide range of results?
I have two VV's, one FC, one NL presented to the fileserver. While no other activity is happening on the array, I copy from one volume to the other and back again numerous times (only one copy job running at any one time).
The results are ranging from 20MB/s up to 200MB/s for a 1.5GB test file.
Can anyone explain why such a wide range of results?
Re: IOPS per Physical Disk?
I've also seen these numbers referenced in the 3PAR OSSA report (Over Subscribed System Alert). In general these are good numbers to go by. The NL drives can handle spikes higher than 75 IOPS but an average of 200 IOPS seems extremely high. You might want to use the Informs Console and setup a new chart to monitor the service times of the NL drives to see how they are handling the load. If they are overloaded the service times will usually indicate this.
As far as a solution, I would first check the ESX guests and verify they have enough RAM assigned. I've ran into a problem recently where several guests were running low on RAM and causing excessive IO due to paging.
I'm a new 3PAR owner trying to learn the ropes myself.
As far as a solution, I would first check the ESX guests and verify they have enough RAM assigned. I've ran into a problem recently where several guests were running low on RAM and causing excessive IO due to paging.
I'm a new 3PAR owner trying to learn the ropes myself.
Re: IOPS per Physical Disk?
Don't confuse IOPS with throughput....
IOPS is the number of IO's a disk (or an array) can handle. Though troughput is related to that it cannot be considered the same.
E.g. a 100GB SQL DB can generate a lot of (quite small) IOPS, but throughput could be low. And a file copy of large files might have quite a througput but the amount of IOPS could be low.
Regards,
Martien
IOPS is the number of IO's a disk (or an array) can handle. Though troughput is related to that it cannot be considered the same.
E.g. a 100GB SQL DB can generate a lot of (quite small) IOPS, but throughput could be low. And a file copy of large files might have quite a througput but the amount of IOPS could be low.
Regards,
Martien
Re: IOPS per Physical Disk?
We also have been fighting an IOPs issue. The previous regime did not take into consideration performance but bought solely based the purchase on capacity. We have rougly 30 VMWARE hosts with 30 shared VVs over 80 NL spindles (2 node) which should give us around 6000 IOPs at manufacturer specs. We consistently stay at 10k IOPs with peaks to 15k. Windows is OK with having a little latency but VMWARE is very touchy. So disks can't read/write fast enough and then creates a backup especially on a 2 node system. Long story short is more spindles, more performance. It can be tough to walk the fine line between capacity, spindle count and what you can afford in your budget.
If you don't have System Reporter, I would suggest it. I works well for pulling all the array stats and gives nice reports that can be exported to Excel and further tweaked.
Good Luck!
If you don't have System Reporter, I would suggest it. I works well for pulling all the array stats and gives nice reports that can be exported to Excel and further tweaked.
Good Luck!
Re: IOPS per Physical Disk?
Now that you understand which Exchange activities and components generate disk I/O and how to configure your storage to support them, you must calculate the disk I/O requirements for your users. Calculating your disk I/O requirements ultimately allows you to optimize your disk subsystem to best support your users.
Your goal is to provide enough disk I/O performance (measured by the number of I/O operations per second [IOPS] that can be performed) with acceptable latency that allows for efficient Exchange functionality.
Calculating the IOPS per mailbox is a convenient way to measure the profile for a given server based on random database read/write I/O (transaction log I/O is not factored into this equation). The higher the IOPS per mailbox, the more aggressive the mailbox profile is in terms of disk usage.
There are two approaches to calculating your disk I/O requirements:
Determine user needs based on theoretical data
Calculate user activity by using the Performance console (Perfmon)...
Your goal is to provide enough disk I/O performance (measured by the number of I/O operations per second [IOPS] that can be performed) with acceptable latency that allows for efficient Exchange functionality.
Calculating the IOPS per mailbox is a convenient way to measure the profile for a given server based on random database read/write I/O (transaction log I/O is not factored into this equation). The higher the IOPS per mailbox, the more aggressive the mailbox profile is in terms of disk usage.
There are two approaches to calculating your disk I/O requirements:
Determine user needs based on theoretical data
Calculate user activity by using the Performance console (Perfmon)...