how replication and snapshot work

[frigomiam]

Hi,

V&B8,

I would like please to clarify the mechanism of replication and snapshot.

example :

I have my replica VM already sitting with 1 restore point on my ESX at the DR.
replication jobs kicks in

the restore point ( meaning 1 snapshot file ) will be deleted ?
the changes gathered from CBT on source will be uploaded to target ESX. Will these changes be injected into base vmdk of the replica vm ?
then a new restore point will be created ?

thanks,
francois

[frigomiam]

I want also to add the retention policy is only 1 restore point.

[foggy]

Francois, each replica restore point is a separate VM snapshot. In your case, after creating new restore point (snapshot), the job will delete the previous one, since it falls out of the retention policy. Corresponding blocks will be injected into the base VMDK.

[frigomiam]

thanks foggy,

I still trying to understand,
restore point is an empty snapshot file, base disk and parent snapshots contain the vm datas.

"job will delete the previous one" > it deletes the snapshot without commiting into the base disk ? ( because I understand you can power on the replica VM manually without requiring new snapshot)

how does it work when you have more than 1 restore point ? it will delete the last vm snapshot without commiting to upper snapshot file, then inject cbt changes into a new vm snapshot file, then create a new snapshot to make it last restore point ?

is there a doc that explains in detail how veeam works with snapshots ?

[foggy]

For some reason you seem to misunderstand my previous reply, sorry if it was confusing, I tried to make it as simple as possible. Each incremental job run creates a new snapshot placing all the changed blocks in it. At the end of the job retention comes into action, deleting the oldest snapshot in the chain (basically, committing its data into the base disk). You can find the link to the corresponding user guide section in my previous post.

[frigomiam]

ok, thanks foggy, may I ask you :

how does veeam handle the fact that the replica VM can be powered on manually ? I understand you don't need to create a new snapshot in this case.

I look now at my replica VM, I see it has 1 snapshot file and the base disk. If I power on the VM, changes will be written in the same snapshot.
What happens then when veeam starts a new replication ?

[foggy]

Restore point snapshot will remain intact, since all disk changes after turning the replica VM on will go into the new snapshot file that is automatically created to protect the selected snapshot's state (standard VMware snapshot tree functionality). This snapshot file will be deleted during the next replication job run and the changes made during the testing phase will be discarded (new restore point snapshot will be created).

[frigomiam]

well, here is what happens for me in reality :

- I have vsphere 5.5 and V&B8, all replication works OK.
- retention policy = 1 so I have only 1 snapshot per VM.
- If I power on 1 replica VM, no new snapshot is created, no new snapshot file in the vmfs.
- I see in the .vmsd file that there is 1 snapshot indeed for 1 restore point.
- the snapshot file is only 84KB which is almost empty. after I run the VM the snapshot file grow.

so it's a bit misleading, here the restore data are already commited to the base disk, and there is indeed 1 new snapshot restore point, but restore data are not contained in the snapshot file.

Code: Select all

example :

/vmfs/volumes/4f686426-ebc5fd19-08f2-0026554f0564/DUB-PRINT-SRV1_replica_1 # grep scsi WIND2k81.vmx
scsi0.present = "true"
scsi0.sharedBus = "none"
scsi0.virtualDev = "lsisas1068"
scsi0.pciSlotNumber = "160"
scsi0.sasWWID = "50 05 05 65 ae d8 5d 20"
scsi0:0.present = "true"
scsi0:0.fileName = "WIND2k81-000002.vmdk"
scsi0:0.deviceType = "scsi-hardDisk"
/vmfs/volumes/4f686426-ebc5fd19-08f2-0026554f0564/DUB-PRINT-SRV1_replica_1 #


DUB-PRINT-SRV1_replica_1 # ls -lh
total 28557952
-rw-------    1 root     root       84.0K Jul  8 22:49 WIND2k81-000002-delta.vmdk
-rw-------    1 root     root         319 Jul  8 23:02 WIND2k81-000002.vmdk
-rw-------    1 root     root       34.1K Jul  8 22:49 WIND2k81-Snapshot365.vmsn
-rw-------    1 root     root       33.8K Dec 31  2014 WIND2k81-Snapshot449.vmsn
-rw-------    1 root     root       33.8K Jan  2  2015 WIND2k81-Snapshot455.vmsn
-rw-------    1 root     root       40.0G Jul  8 23:02 WIND2k81-flat.vmdk
-rw-------    1 root     root        8.5K Jul  8 22:49 WIND2k81.nvram
-rw-------    1 root     root         518 Jul  8 23:02 WIND2k81.vmdk
-rw-------    1 root     root         589 Jul  8 23:02 WIND2k81.vmsd
-rwx------    1 root     root        5.7K Jul  8 23:02 WIND2k81.vmx
-rw-------    1 root     root        3.7K Jul  8 22:49 WIND2k81.vmxf
-rw-------    1 root     root         843 Jan  2  2015 WIND2k81_oldvmsd
-rw-------    1 root     root      217.8K Jan 21 16:10 vmware-11.log
-rw-------    1 root     root      217.2K Jan 26 18:09 vmware-12.log
-rw-------    1 root     root      217.0K Feb  4 17:09 vmware-13.log
-rw-------    1 root     root      217.0K Feb 21 13:18 vmware-14.log
-rw-------    1 root     root      219.8K Mar 20 10:36 vmware-15.log
-rw-------    1 root     root      217.1K Mar 21 12:39 vmware-16.log
-rw-------    1 root     root      217.2K Jun 19 10:21 vmware.log
/vmfs/volumes/4f686426-ebc5fd19-08f2-0026554f0564/DUB-PRINT-SRV1_replica_1 # cat WIND2k81.vmsd
.encoding = "UTF-8"
snapshot.lastUID = "365"
snapshot.current = "365"
snapshot0.uid = "365"
snapshot0.filename = "WIND2k81-Snapshot365.vmsn"
snapshot0.displayName = "Restore Point 08-07-2015 23:44:50"
snapshot0.description = "<RPData PointTime=|225247405941336944624|22 WorkingSnapshotTime=|225247405941651629375|22 PointSize=|221898346572|22 PointType=|22ECompleteRestorePoint|22 />"
snapshot0.createTimeHigh = "334436"
snapshot0.createTimeLow = "-215885488"
snapshot0.numDisks = "1"
snapshot0.disk0.fileName = "WIND2k81.vmdk"
snapshot0.disk0.node = "scsi0:0"
snapshot.numSnapshots = "1"
/vmfs/volumes/4f686426-ebc5fd19-08f2-0026554f0564/DUB-PRINT-SRV1_replica_

[foggy]

Right, that empty snapshot file is the protective one, in which changes are written on starting the replica VM. VM data are stored in the base disk, since you have just 1 point retention. My previous comment was more for a general case of several restore points, which are stored as snapshots.