2017-04-13 03:27 AM - edited 2017-04-13 03:29 AM
Let's imagine a situation that there is a 8000 system cluster-in-box with single flash disks shelf - 11 disks are provisioned for one node and 11 - for second with root-data provisioning. just one aggregate per node with root and data volumes.
Now, 2/4/6 new shelves come with 10K SAS drives. Can they be added directly to the current configuration by extending the existing aggregate and having flash disks acting as cache for SAS disks or complete reconfiguration/clean-up of the system is required?
Anyone has gone thru the similar experience?
Solved! SEE THE SOLUTION
2017-04-13 04:52 AM
If the 8000 series system you're talking about is an AFF8xxx then adding HDDs to the system is a non-starter:
"Only SSD drives are supported on AFF platforms. If non-SSD drives are attached to an AFF platform, the disks do not go online."
However, if it's just a FAS8xxx that shipped only with a SSD shelf, then assuming all the SSD partitions have already been allocated to aggregates you would have to destroy those aggregates to build what you're talking about (i.e. a FlashPool/hybrid aggregate). You'll need free SSD partitions to either build a Storage Pool (which then gets allocated to the HDD aggregate) or partitions you'll add directly to the HDD aggregate.
ONTAP 9 documentation for FlashPool aggregates: http://docs.netapp.com/ontap-9/index.jsp?topic=%2Fcom.netapp.doc.onc-sm-help%2FGUID-2E861450-119C-4793-B4C9-0B94577E43C6.html
Hope that helps,
2017-04-13 05:07 AM
Assuming the system was purchased as an AFF then the answer is no to any option.
The second part of the question really depends on if all of the SSD partitions were allocated to root/data aggregates. If you have SSD partitions available (i.e. unallocated) from the ADPv2 disk pool then you can create a storage pool and/or add those partitions to a HDD aggregate. But if all the SSD partitions are otherwise allocated then they're not available to be added to a hybrid aggregate.
Hope that helps,
2017-04-13 05:22 AM
Aha, that's interesting. Is there a way I can control the size of partitions on flash disks?
So let's imagine that we have 24 flash disks, 12 for each node. How would I configure these disks by creting smallest root and small data partition for each node and then allocate the rest of the available disk for flash pool?
Let's image that we have 600GB flash drive (size does not matter), 11 of such for each node, 9 usable with DP. What would be minimum size of root and data partitions created on that aggregate so I can use unused portion for flash pool? Moreover - can only, smallest size root partition created on that aggregate and the rest of disk used for flash pool?
2017-04-13 05:29 AM
root-data(-data) sizes are determined automatically during intitial setup and cannot be changed afterwards. I am not aware of any method to influence their sizes. You cannot mix usage - either disk is used for root-data or for pool.