MonthMarch 2012

Netapp Volume Migration

As a storage admin , volume migration is always an active routine that need to go through every months or quarters. I was involving in most storage migration using lots of method for the past few years and here is my finding.

  1. Always use snapmirror for volume migration! This is the best way to migrate data in the most consistent way.
  2. Avoid using rsync for huge volume with multiple sessions(directory too huge and you have to split it to multiple sessions).
  3. Detailed plan and down time needed.
Let me share with you my own best practice:
  1. Source filer
    • Identify Volume
    • Volume Size
    • Aggregate type (32bit or 64bit)
  2. Destination filer
    • Identify new volume (with new volume name)
    • New volume size (must be greater than source volume during transfer to minimize the chances of failing during transfer. snapshot might take up additional space on the migration)
    • Aggregate balanced space . Aggregate free space is important for performance , best recommended free space of 15%-20%.
    • Aggregate type (32bit or 63bit)
  3. Getting downtime
    • Engage with customer to get the downtime. Usually the downtime i request is less than 2 hours as the final sync is usually less than 10minutes because i have scheduled snapmirror running daily for the incremental changes.

  4. Cut over plan
    • Always schedule snapmirror schedule to update daily during off hour(or when the filer is free).
    • Run manual snapmirror update for few time an hour before the final sync.
    • restrict source volume after the migration to prevent new data writing in.
    • Test the access in new destination (NFS/CIFS/iSCSI)

Snapmirror alert script

Snapmirror is very cool and fantastic tool command used for migration and mirroring. You can run the snapmirror using schedule (snapmirror.conf) and let it run.

However, the draw back of snapmirror is

  • keep the information for last transfer only. No more history more than last transfer.
  • NO Alert sent when the job fail due to destination space is full (& others reason) .

 
Continue reading

© 2018 Thinkway

Theme by Anders NorénUp ↑