If I move a member from one storage pool to another will there be any down time with the volumes hosted by the member I am moving? My goal is to move both members to the same storage pool to take advantage of tiering. Which brings me to my next question.
Will there be any issues running a ps6510(48x 10K SAS RAID50) and a ps6110xs(7x SSD 17x 10K SAS RAID-6 Accelerated) in the same storage pool? It seems from all the reading that as long as we are using different RAID levels on the members since there are mixed drives we should be good to go....
Regarding moving a member to a new pool:
When a member is moved to a new pool, no data is moved with it. The remaining members will host the volumes contained from the member you move. So it is vitally important to ensure that the remaining members have enough free space (plus 5% for array operations) to host all the data from the member moving to the new pool.
Once the member is in the new pool, you can then re-assign the volumes to the new pool.
Regarding downtime during this operation, this can be done during “live data I/O” however depending on the amount of data and I/O this can take anywhere from 24 hours to a several days to complete. This will be a background process, but during this time group performance could be somewhat degraded, so plan accordingly. The group admin guide discusses this in more detail.
Regarding mixing the two Array types in the same pool:
It is possible to mix these two types of array in the same pool, but if you do put them in the same pool, the performance balancing will tend to move hot pages to the less busy array. There is a latency difference threshold that will determine when that happens. However, the existing 6510 is larger by a fair amount compared to the 6110xs so a larger proportion of the volume slices will reside on this member, so it will end up doing most of the work.
The hybid arrays (SSD and SAS in the same array) use their own APLB for unbalanced loads on that member. The APLB is able to work in conjunction with the hybrid arrays, permitting other, non-hybrid arrays to migrate active data to the hybrids in exchange for inactive data. If there are more active data is placed on a hybrid than the SSD tier is able to hold, the APLB is able to migrate active data from the HDD tier to another member such as a second hybrid or standard array that may have more resources available to better service the I/O requests.
In your case, you also have two different RAID policies, so unless you set the volumes to a specific RAID Type, the volumes will (after time) be load balanced automatically to the member that better matches the RAID characteristics of the data I/O.
Also consider the iSCSI connections, more pools equals more connections, but this may not be issue if you are below the max amount now (1024 per pool, 4096 up to 4 pools)
If you haven’t seen this, review the “Dell EqualLogic PS Series Architecture: Load Balancer” which discusses this in greater detail: www.equallogic.com/.../DownloadAsset.aspx
Social Media and Community Professional#IWork4DellGet Support on Twitter - @dellcarespro
Follow me on Twitter: @joesatdell
Thanks for the info!
I just thought about your above answer and I have a further question. In my case the members I am moving to new pools are in single member pools. With that being the case there would be no remaining members to host the volumes. So what happens in that instance?
Single Member in Default Pool:
It is not possible to move the member to another pool if the single member is in the default pool. You would need to have another member in the default pool in order to hold the data of the member you are moving. To visualize this, as a test (safe to do on live data): if you create the new pool, then select the Member (the single member in the default pool) in the GUI, and select the “Modify member setting” and try to assign the member to a new pool (under the "storage pool assignment" section); you will see an error that there isn’t enough free space in the current pool to vacate the member.
The work-a-rounds are:
A) add a temporary member to the default pool to hold all the data of the one you are moving
B) back-up or copy the data off of the volumes, delete all the volumes from the pool, then you can move the member, and restore the data
C) move the volumes in the pool to another member in a different pool.
Single Member in a Non-Default Pool (also works with multi-member pools):
If the member is not in the default pool, you can merge any storage pool into another pool, called the destination pool (but you can’t merge to the default pool).
Merging a pool into a destination pool moves the pool members and volumes data into the destination pool. The group then deletes the empty pool.