What does "lag" mean for SnapMirror/SnapVault?

Applies to

ONTAP 9

Answer

• A SnapMirror or SnapVault relationship's lag time is calculated using:
  • The snapshot timestamp
  • The time on the destination system
  • The amount of time needed to transfer the snapshot from source to destination

• The term 'lag' is typically associated with performance, with the common perception that lag is the elapsed time since the last successful update
• While this is not completely incorrect, it does not account for 2 other factors:
  • The time, based on the clock and timezone, on the source and destination storage controllers
  • The duration of the transfer

• The time on the source and destination is important because this determines the timestamps on the file system and snapshots.
  • If time is configured incorrectly, timestamps will be inaccurate  
  • Because lag is calculated based on snapshot timestamps, if the time is not correct, lag will not be correct  

• The duration of the transfer is also overlooked because of the nature of replication
  • Lag is not measured based only on the time a transfer starts and completes
  • Lag is measured from the time the snapshot is created on the source, plus the duration of the transfer. 
    • Transfer can be schedule update or manual update transfers. 

• Consider the following SnapMirror scenario:

  Source	Destination
  ControllerA:vol_1	ControllerB:vol_1_mir

1. A scheduled update starts at 12:00pm
2. A SnapMirror snapshot is created on the Source volume, and a transfer is started 
3. The transfer takes 45 minutes to complete
4. The time on the destination system is now 12:46pm
5. The transfer completed 1 minute ago

If measured during step 5, the lag is 46 minutes, because:

• 46 minutes have elapsed since the snapshot was created on the source
• 46 minutes elapsed since the snapshot was successfully transferred to the destination

• On the Destination, Lag is calculated by finding the difference between:
  • The snapshot creation timestamp
  • The time on the destination, based on the destination storage controller's clock 
  • If the time is not configured correctly on the destination or source, the lag time will be incorrect

• Consider the following scenario:

  Primary	Secondary
  CIFS_SVM:vol_1	CIFS_DR:vol_1_dr

1. Based on the snapshot policy on vol_1, a snapshot is created at 5pm
2. The snapshot is created with the snapmirror-label sv_daily
3. At 1 AM the following morning, a scheduled snapmirror update is triggered, configured to replicate any snapshot labeled sv_daily
4. The transfer takes 30 minutes to complete

The lag in this scenario would be 8 hours 30 minutes, because:

• At the time of the scheduled snapmirror update, eight hours had elapsed since the snapshot was created and labeled sv_daily
• Transferring the snapshot from the source to the destination took 30 minutes

In summary

• Lag is the difference between the snapshot timestamp and the time on the destination system
• Lag includes the amount of time needed to transfer a snapshot from source to destination
• When examined in the context of snapshot timestamp, and the duration of the transfer, "long" lag times are often found to be normal

Additional Information