- ONTAP 9
- Clustered Data Ontap 8
Why does SnapVault take so long
- SnapVault backup involves starting the baseline transfers, making scheduled incremental transfers.
- Baseline transfers are the longest.
- SnapVault is a background process and will throttle back by default with increased user workload on source or destination nodes.
Phases of SnapVault Initialization
- Phase 1: SnapVault will initialize rapid transfer speed and good network throughput for the first stage. Directory, file metadata,Access Control List (ACLs), and inode information are transferred to a holding station on the destination volume. This area should not be visible to clients.
- Phase 2: An extended period of zero data transfer occurs.The only network operations seen during this stage are keep alive and status messages that are sent between filers. On the destination, the file structure is built in the qtree using the information from Phase 1. Directories with lots of small files will slow this phase down. This phase can be very long.
- Phase 3: Good throughput occurs as the file data is rapidly transferred and written into the directory structure in the qtree that exists on the destination.
Why does it look like there is no data is being transferred during SnapVault
Refer back to the Phases of Initialization, if in Phase 2 this phase does not have data transfer while setting up the file structure.
Ways to improve SnapVault performance
- Spread out files in multiple directories.
- Increase the number of disks per raid group. (spindle count)
- Disk types that have faster Revolution Per Minute(RPMS) (FCAL, SATA or SAS) can improve performance.
- The greater the number of files/inodes, the longer the inode processing will take during initialization and updates. Its not possible to speed up this processing