NetBackup™ Deduplication Guide

Snapshot retention:

• "cloudFullTaskInterval": 36000,: Automatically creates the full snapshot for the universal share interval and the default value is 10 hours. This entry must be an integer using the unit of seconds.

• "cloudIncrTaskInterval": 1800,: Automatically creates the incremental snapshot for the universal share interval and the default value is 30 minutes. This entry must be an integer using the unit of seconds.

• "cloudFullSnapshotRetention": 172800,: The retention time of the full snapshot copy. When the retention expires, the full snapshot is deleted from local storage and the cloud bucket storage. The default value is 48 hours. If the retention is set longer than 48 hours, there might be an effect on space reclamation.

Local disk cache configuration:

• "CloudCacheSize": 500,: The local disk cache size for the universal share and instant access. This option applies only to the universal share with object store and instant access with object store. The vpfsd command removes this amount of space from the spoold service so you must verify there is enough space for the cache size. Otherwise, the universal share with object store or instant access for the object store are not created. MSDP verifies that there is enough configured space before the creation of the universal share. You must restart vpfsd when you increase the size of the cache and be aware that if there is not enough free space for the cache size, vpfsd cannot start.

  After the universal share or instant access with object store is removed, that amount of space does not return to spoold automatically. Reduce the CloudCacheSize to return some space back to spoold. You must restart vpfsd after the removal.

• "CloudCacheLowThreshold": 50,: The vpfsd service starts to reclaim the space of the cloud cache when the space usage of the cache reaches the low threshold. This entry is in the unit of percentage.

• "CloudCacheHighThreshold": 85,: The vpfsd service stops any data from being written or the download of any data when the space usage of the cache reaches the high threshold. The data write and the data download continue when there is some free space. This entry is in the unit of percentage.

Local disk cache sizing:

• The size of the cache depends on number of the data volumes, number of the universal shares, and the total number of the files in all of the universal shares. There are two types of data stores in the local disk cache: the metadata and the deduplicated segments in the data containers. A greater number of files requires more disk cache, and a greater number of data volumes also requires more disk cache.

  Each one million files requires about 300 GB disk cache for metadata of universal share. For the data container cache, each universal share has 32 data container caches for each data volume by default, and there are 1024 data containers for the read cache by default for all the universal shares.

  The following is the algorithm to calculate the disk cache size:

  Metadata size: 300 GB * Number of File (unit is million)

  Data container size: 64 MB * (32 * Number of data volume * number of share + 1024)

  Total cache size: (metadata size + data container size) / 0.85 (CloudCacheHighThreshold)

  The default size of the local disk is 500 GB. This size can handle most scenarios. The configuration below is the recommend setting if the number of data volumes or partition is 4 or less. Configure CloudCacheSize in vpfsd_config.json to tune the cache size.

  Number of files (in millions)	Number of shares	Number of partitions	Disk cache size (in GBs)
  0.5	1	4	400
  0.5	20	4	400
  1	1	4	600
  1	20	4	600
  2	1	4	800
  3	1	4	1200
  4	1	4	1600
  5	1	4	2000
  5	20	4	2000
  6	1	4	2200
  7	1	4	2600
  8	1	4	3000
  9	1	4	3400
  10	1	4	3800
  10	20	4	3800

Snapshot management:

• List all of the snapshots which include the full snapshot and incremental snapshot in the cloud bucket:

  /usr/openv/pdde/vpfs/bin/vpfscld  - list

• Manually take snapshot and upload snapshot and data to cloud bucket:

  /usr/openv/pdde/vpfs/bin/vpfscld --snapshot 
  --share_id <share> --snap_type <full|incr>

• Manually remove a snapshot from local and cloud, please be aware of that the expired snapshot is not recoverable:

  /usr/openv/pdde/vpfs/bin/vpfscld --expire 
  --share_id <share> --pit <point in time>

• Manually recover a snapshot from a cloud bucket:

  /usr/openv/pdde/vpfs/bin/vpfscld -recover 
  --share_id <share> [--tgt_id <target>] [--pit <point in time>]
  [--force]

• Configure the number of snapshot copies for universal share.

  The default number of snapshot copies is 2. It can be configured with the cloudMaxSnapshotCopy parameter in the vpfsd_config.json. When the value of cloudMaxSnapshotCopy set to 0, it disables the feature that automatically takes the snapshot. Higher values require more local disk cache if the universal share stores the snapshot on local disk cache.

• Configure universal share stores snapshot copies on the local disk cache.

  With NetBackup 10.4 and later versions, snapshots will be stored only on the cloud bucket instead of being stored on the local disk cache. However, you can configure and control the storage of additional snapshots on the local disk cache. If you still want to keep the snapshot on the local disk cache, you must configure cloudKeepLocalSnapshotCopy as true in vpfsd_config.json. It also requires more local disk space.