Skip to main content
commvault.com commvault.com

Hi Team,

 

We’re discussing this internally within our organisation and wish to gain some clarification how we perform I/O against NFS object storage.

 

Kind regards,

Jon

 

Hi Jon 

 

By design for NFS ObjectStore the creation/modification/deletion of files function as below:

 

Write Process

  • Client writes a File to the NFS Share (single Write), it is written to NFS cache in the form of 10 MB blocks (called as extents).

  • The metadata about the file and extents is written to a journal file.

  • The metadata sits in Journal File for up to a minute and then flushed to the Index Server (Index Cloud).

  • So index cloud holds the information about the files/folders and extents.

 

Read Process:

  • During a read operation, the user will first access the file from the NFS Share.

  • If the required extent exist in the NFS Cache it will continue to read from the Cache, otherwise it will reach out to the Index Server and replay the extents

 

File Delete and Prune Process:

  • We will first update the Index (on the Index Server) and mark the file as deleted.

  • As per the retention criteria, the admin process (pruning thread which runs every 24 hours) will mark the file as pruned.

 

Kind Regards

 

WW

Thanks @Winston W, that answers it perfectly!

 

Take care, ta!

Reply


Terms & ConditionsCookie settings