sfcredfox
Patron
- Joined
- Aug 26, 2014
- Messages
- 340
Community,
I'm looking for any first hand experience with performance best practices and recommended configurations for doing FibreChannel connected FreeNAS storage for use as VMware Data Stores.
This post is an excellent configuration guide for enabling FC:
https://forums.freenas.org/index.php?threads/freenas-9-3-fc-fibre-channel-target-mode-das-san.27653/
This guide worked well back when I implemented it.
As many VMware folks know, there's configuration and setup documentation from VMware, and then more specific performance related best practices from every storage vender in terms of how to best setup MPIO, queue depth, recommended datastore sizing for specific or general workloads, recommended VM counts per datastore, etc.
We've been running a dual fabric FC setup for a while now for our VMware lab, and I'm interested in what other people have come up with in terms of their performance testing.
Examples:
I have two disk pools VMware: (1) 24 disk 3G SAS pool and (2) 12 disk 6G sas pool. We created two 1TB datastores on each of them.
Lots of storage venders have different ideas/testing on creating less datastores that are larger, versus more datastores that are smaller. queue depths come into play with these models.
Also, when doing MPIO (VMware's Round Robin specifically), under certain circumstances, adjusting the pending IOPS from default (1000) to a smaller number (most lab people do 1) can result in performance increases, but that's not a hard/fast rule and depends on how many VMs you have and other work load specifics. Again, a storage vender normally has done all that performance testing and has a list of recommendations.
Lots of blogs out there and small lab posts talk about changing some of these things, but there doesn't appear to be much in the way of quantifiable results. You don't need EMC or NetApp to quantify their recommendations unless your case is very unique, but FreeNAS community side doesn't have this, so I'm interested in what people have found.
Questions:
1. Does anyone have any first hand experience with doing some of this performance testing, best practice, datastore sizes, etc for doing this with FreeNAS?
2. What results did you get?
Thanks
I'm looking for any first hand experience with performance best practices and recommended configurations for doing FibreChannel connected FreeNAS storage for use as VMware Data Stores.
This post is an excellent configuration guide for enabling FC:
https://forums.freenas.org/index.php?threads/freenas-9-3-fc-fibre-channel-target-mode-das-san.27653/
This guide worked well back when I implemented it.
As many VMware folks know, there's configuration and setup documentation from VMware, and then more specific performance related best practices from every storage vender in terms of how to best setup MPIO, queue depth, recommended datastore sizing for specific or general workloads, recommended VM counts per datastore, etc.
We've been running a dual fabric FC setup for a while now for our VMware lab, and I'm interested in what other people have come up with in terms of their performance testing.
Examples:
I have two disk pools VMware: (1) 24 disk 3G SAS pool and (2) 12 disk 6G sas pool. We created two 1TB datastores on each of them.
Lots of storage venders have different ideas/testing on creating less datastores that are larger, versus more datastores that are smaller. queue depths come into play with these models.
Also, when doing MPIO (VMware's Round Robin specifically), under certain circumstances, adjusting the pending IOPS from default (1000) to a smaller number (most lab people do 1) can result in performance increases, but that's not a hard/fast rule and depends on how many VMs you have and other work load specifics. Again, a storage vender normally has done all that performance testing and has a list of recommendations.
Lots of blogs out there and small lab posts talk about changing some of these things, but there doesn't appear to be much in the way of quantifiable results. You don't need EMC or NetApp to quantify their recommendations unless your case is very unique, but FreeNAS community side doesn't have this, so I'm interested in what people have found.
Questions:
1. Does anyone have any first hand experience with doing some of this performance testing, best practice, datastore sizes, etc for doing this with FreeNAS?
2. What results did you get?
Thanks
