if you change the mirroring direction volume B becomes primary and then a host attached to volume B can read from and write data on it.
That's one of the main-features for the metro/global mirroring functions, to get a fully functional copy of the production data on a remote site.
But before you switch the direction take care that the host attachted to volume A is not using the volume anymore (shutdown). After a Moment you will see a message for the copy relationship that the copies are "in sync). I think for global mirror you get also a timestamp. So if host A is shutdown or detached and the copy relation is in sync then you can switch the direction delete the copy relationship. Volume B becomes a normal volume and is ready for usage (read and write).
https://www.ibm.com/docs/en/flashsystem-7x00/8.5.x?topic=functions-remote-copy-function
https://www.ibm.com/docs/en/flashsystem-7x00/8.5.x?topic=administering-managing-copy-services
These two topic are related to the copy-services on a Flashsystem or Storwize (the commands are the same)
Original Message:
Sent: Wed March 22, 2023 07:09 AM
From: Adrian Nicolae
Subject: Volume size on FCM3 Standard Pools
I understand the remote copy flow. But :
- volume A has a 500GB allocated size , real capacity = 72GB, used capacity = 62GB ( the source being a 2076-624 with RTC compression and no dedup)
- the target volume B after 100% progress for the rc relationship has the same allocated size of 500GB on 7300 (but effective capacity not usable if I'm understood well) but real capacity = 500GB and used capacity = 500GB. So it looks full even if the GMCV relationshp doesn't have any error. Maybe there are a lot of untrimmed blocks copied from source to target.
So what happens when the volume B will be primary ? Is the host able to write new data or not ? The host will still see that it has enough free storage from its point of view, the written data is the same.
------------------------------
Adrian Nicolae
Original Message:
Sent: Wed March 22, 2023 06:57 AM
From: Dorde Knezevic
Subject: Volume size on FCM3 Standard Pools
OK, lets say the source volume on the old system is "A" and the target-volume on the 7300 is "B".
When the mirroring is finished/in sync, then the content on A and B is identical. And A or B will appear identical to the attached OS (except the basic ID's of the 7300/volume).
So if the mirror is in sync you can switch to the Volume B an use it like you did it with Volume A. The Difference between DRP and FCM is the way the data is compressed and which methods will be used. Data consistency has the highest priority, so all of data reduction features have to work in the background and the content of the volume must be "untouched".
------------------------------
Dorde Knezevic
Original Message:
Sent: Wed March 22, 2023 05:49 AM
From: Adrian Nicolae
Subject: Volume size on FCM3 Standard Pools
>>When you copy a 100 GB volume via Metro or Global Mirror then the copy job is done between 2 different systems and all block's on the volume will be copied, also the unused space. That's normal behavior.
This means that the host can write new data to the target volume after switching the replication even if it's full because it's including unused space (untrimmed data maybe) ? Or is it safer to extend the master and replicate the volume again ?
------------------------------
Adrian Nicolae
Original Message:
Sent: Wed March 22, 2023 04:33 AM
From: Dorde Knezevic
Subject: Volume size on FCM3 Standard Pools
Hello Adrian,
the volume size on the 7300 must be the same like on the source system. In other words the host expects 100 GB and the 7300 delivers 100 GB. But in the background the FCM's are compressing the data. So for the 100 GB host volume you need just 50 GB on the FCM's. That's how it works and that's fine.
When you copy a 100 GB volume via Metro or Global Mirror then the copy job is done between 2 different systems and all block's on the volume will be copied, also the unused space. That's normal behavior. The compression on the 7300 will reduce the usage on the pools as much as possible.
Regards,
Dorde Knezevic
------------------------------
Dorde Knezevic
Original Message:
Sent: Fri March 17, 2023 01:53 PM
From: Adrian Nicolae
Subject: Volume size on FCM3 Standard Pools
Hi,
I'm migrating some volumes from older controllers (2076-624 with RTC compression and 2076-724 with DRP) to a brand new 7300 with FCM3-only drives and I'm a little puzzled about the volume size/used figures on the new FCM system.
My issue is like this :
- on 7300 I've created a standard pool , I want to get rid of all the DRP stuff and use only the drive compression
- the 7300 volumes are thin
- I've replicated a few volumes from the older systems to the new system with GMCV or GM
- on the older systems the volumes were around 50% used. The volume size is strictly corelated with the used mdisk space after compression for that volume
- after replication I'm surprised to see that the auxiliary volumes are 100% full on the 7300 system. Both the master and the aux have the same volume size of course and they are in sync. However, the used volume size is 50% on the older system and 100% on the new system.
- on the FCM system the drive compression is working fine at the pool level. I have a compression ratio greater than 2:1 so the actual usable storage is as expected but I'm puzzled about the volume size/usage concepts on the 7300 system considering that I'm already getting thin provisioned volume copy space warnings :)
- my guess is that on the FCM pool the volume size and the volume usage is actually considerred based on the max written data and not on the actually used mdisk space as on the older system. Is this correct or am I doing something wrong ?
Thanks.
------------------------------
Adrian Nicolae
------------------------------