Current Environment : HFC (High File Count NAS Environment ) , 10 node Cluster Ontap9.1 , NDMP , dozens of LTO Tape Drives.
What can we do before :
1. Production Cluster hourly snapmirror update to DR Cluster ( DP snapmirror )
2. Monthly Full NDMP Backup for every Volume ; Daily NDMP Incremental backup every 3~5 days.
3. Source Volume take snapshot every 12hr and keep 4 copies.
What we can do now :
1. Keep 1hr RPO ( hourly XDP snapmirror update )
2. Keep 30 Daily snapshot at DR Cluster as Daily Backup
3. Source Volume take snapshot every 12hr and keep 4 copies.
4. Only Full NDMP backup using Tape Drive
My procedure as below :
1. create new snapmirror policy name "Vault_Daily" as tpye "mirror-vault"
2. add-rule to policy "Vault_Daily" for Destination Local take snapshot daily
snapmirror policy add-rule -vserver svm1 -policy Vault_Daily -schedule daily -snapmirror-label N_Daily -keep 30
PS : This rule is used to create Daily snapshot ONLY at destination cluster Local only ( not update and transfer from source )
Then we can see destination volume have daily snapshot named as N_Daily...
3.
For new volume create XDP snapmirror relationship and using policy "Vault_Daily" then initialize.
For original DP relateionship volumes, you can take links below to convert DP to XDP snapmirror .
PS: make sure all destination volume is thin-provition and snapshot autodelete is disable.
4. Enable Volume Efficiency at destination volume for space saving to keep more snapshot copy
volume efficiency modify -compression true -inline-compression true -data-compaction true volumeXX
5. start volume efficiency scan
volume efficiency start -dedupe true -compression true -compaction true -scan-old-data true -b true volumeXX
note1 : -scan-old-data is needed for getting Full saving rate; -b true for compress snapshot data
note2 : (Limitation) the concurrent efficiency jobs is 8 per-node, so it took me several weeks for "-scan-old-data" efficiency jobs for all destination volumes.
PS : in my experience, it scans about 2TB~3TB data every 24hr when scan-old-data. That means if the volume size is 10TB, it may take about 5 days for volume efficiency start with scan-old-data.
note3 : For XDP destination volume that enable volume efficiency, efficiency job can not scheduled, efficiency job will auto start after every snapmirror update.
note4 : During efficiency start as scan-old-data job running, all snapshots created is huge and have no saving, please keep monitor the destination volume space used during scan-old-data. You need to wait for those "giant snapshot" being rorated for release the space used by "no saving giant snapshot".
Then a month after, I am really amazing.
The total volume size of source volumes is around 365388GB. But the total physical size used in Destination Cluster is 301660( include 1hr RPO copy and 30 daily snapshot copies).
I got greate saving rate and really amazing that destination volume with 30 daily snapshot take less space than source volume.
Using XDP snapmirror can meet DR/Backup requirement and working much better than "Synthetic Backup" with any backup software.
