ONTAP Discussions
ONTAP Discussions
Hi.
I work in an public university here in Brazil. We have brought an FAS2750 last year. This unit hve 12x 'standard' disks with 900Gb each. We have one aggregate with size of 6.13Tb and about 11 volumes in this aggregate.
Now we have brought 6x 1.6Tb SSD disks and I'm planning to install in the next 15 days. What I would like to ask is:
1) If I create a new aggregate (5x disks + 1x spare) using RAIDA DP, can I use part (lets say 1Tb) of this new aggregation as flashpool for my current aggreate (non-ssd)?
2) If the answer for the first question is 'yes', it's possible to add this new flashpool to my current aggreate without loosing data? We are using this aggretation in production.
3) Where can I find information/tutoriais do to itens 1 and 2? I would like to install and create this new aggregate myself.
Thank you!
Solved! See The Solution
Hi. You cannot allocate a determined size form an aggregate (for my two years outdated FlashPool knowledge). But you should be able to partition the new disks with ADP, see second paragraph in chapter 5:
https://kb.netapp.com/Advice_and_Troubleshooting/Flash_Storage/AFF_Series/How_to_migrate_to_Advanced_Disk_Partitioning_(ADP) (There's maybe other ways to do it that I'm not aware of - most other NetApp docs refers to NetApp support for the manual partition process, perhaps open a support ticket with your plan ahead of executing your plan).
And add some of these partitions to your existing aggregate:
According to the optimal cache size you find here:
I sometime find that these results conclude that there's no real improvement from Flash Pool (mainly with systems that have Flash Cache). In this case (and in general) it's worth considering putting only intensive or performance-sensitive workloads on dedicated flash aggregate (for example TEMP DB, Indexes, VDI shared drives etc)...
Hi. You cannot allocate a determined size form an aggregate (for my two years outdated FlashPool knowledge). But you should be able to partition the new disks with ADP, see second paragraph in chapter 5:
https://kb.netapp.com/Advice_and_Troubleshooting/Flash_Storage/AFF_Series/How_to_migrate_to_Advanced_Disk_Partitioning_(ADP) (There's maybe other ways to do it that I'm not aware of - most other NetApp docs refers to NetApp support for the manual partition process, perhaps open a support ticket with your plan ahead of executing your plan).
And add some of these partitions to your existing aggregate:
According to the optimal cache size you find here:
I sometime find that these results conclude that there's no real improvement from Flash Pool (mainly with systems that have Flash Cache). In this case (and in general) it's worth considering putting only intensive or performance-sensitive workloads on dedicated flash aggregate (for example TEMP DB, Indexes, VDI shared drives etc)...
Hi Jonis,
As @GidonMarcus mentioned, you cannot specify a part of an aggregate for Flash Pool usage. Once you designate the SSDs are to be used for a Flash Pool aggregate by creating a Storage Pool with the SSDs, it will partition the SSDs into four equal partitions. The operation will then add the partitions to a new SSD only RAID group to the data aggregate you are converting into a Flash Pool aggregate.
This operation is irreversible, meaning you cannot remove the SSD partitions from the data aggregate afterward without migrating all data off of the aggregate and re-creating the aggregate. I highly recommend you make sure Flash Pool is the correct solution for your needs before proceeding or contact your Account Team so they can assess your workload and provide the proper guidance.
Here is a really good read on Flash Pool:
TR-4070: Flash Pool Design and Implementation Guide
Thank you both!
Maybe I'll just use as new aggregate and put most intensive workload there.