cancel
Showing results for 
Search instead for 
Did you mean: 

how does new coming data flows when DCM resync is in progress

david_racoon
Not applicable

Hi, All

I have some questions about VVR DCM resync,

When DCM resync is in progress, and applications still writing new data,

what does VVR need to do with the new coming data ? Is it writing to DCM or SRL ?

and if new data block is conflict with data block already marked in DCM, how does VVR resolve the conflict ?

 

Thanks

2 ACCEPTED SOLUTIONS

Accepted Solutions

mikebounds
Level 6
Partner Accredited

Historically (version 4.x and below), just the DCM would be used for new writes during DCM resync and if a block is already dirty, then it just stays dirty - there is no conflict as DCM is just a bitmap that represents regions on volumes - it does not contain any data.  In recent versions of VVR (5.x onwards), I believe VVR initially uses DCM, but in the latter stages uses the SRL - this is because if you are writing a lot when doing a DCM resync then the DCM resync never finishes, so at some point the SRL is used and when it has cleared the DCM then just the SRL is used as writes written during clearing the DCM will be in the SRL.

Mike

View solution in original post

Douglas_Snyder
Level 5
Employee Accredited Certified

Hello,

VVR handles network failures gracefully with out any disruption to production operations.  In the event network connection between primary and secondary is disrupted VVR will preserves the application writes that are happening at the primary site in the SRL(Storage replicator Log) volume. When the network connection is restored, outstanding writes in the SRL volume will be applied to secondary volumes. The space needed for SRL volume depends on the extent of time the network disruption exists. Space for SRL volume can be extended online. (dynamically with out causing downtime to primary production servers).

In async replication mode, application writes are applied to primary data volume and saved in log volume. VVR will replicate the outstanding writes in the log volume to the secondary utilizing the available bandwidth. Application writes to primary data volume happen independent of replication speed. Writes to secondary data volume continue at the speed determined by the network performance with out effecting the application performance on primary.

Doug Snyder, CISSP

View solution in original post

2 REPLIES 2

mikebounds
Level 6
Partner Accredited

Historically (version 4.x and below), just the DCM would be used for new writes during DCM resync and if a block is already dirty, then it just stays dirty - there is no conflict as DCM is just a bitmap that represents regions on volumes - it does not contain any data.  In recent versions of VVR (5.x onwards), I believe VVR initially uses DCM, but in the latter stages uses the SRL - this is because if you are writing a lot when doing a DCM resync then the DCM resync never finishes, so at some point the SRL is used and when it has cleared the DCM then just the SRL is used as writes written during clearing the DCM will be in the SRL.

Mike

Douglas_Snyder
Level 5
Employee Accredited Certified

Hello,

VVR handles network failures gracefully with out any disruption to production operations.  In the event network connection between primary and secondary is disrupted VVR will preserves the application writes that are happening at the primary site in the SRL(Storage replicator Log) volume. When the network connection is restored, outstanding writes in the SRL volume will be applied to secondary volumes. The space needed for SRL volume depends on the extent of time the network disruption exists. Space for SRL volume can be extended online. (dynamically with out causing downtime to primary production servers).

In async replication mode, application writes are applied to primary data volume and saved in log volume. VVR will replicate the outstanding writes in the log volume to the secondary utilizing the available bandwidth. Application writes to primary data volume happen independent of replication speed. Writes to secondary data volume continue at the speed determined by the network performance with out effecting the application performance on primary.

Doug Snyder, CISSP