In RP, with storage based splitter when the IO comes, it is split up in to 2 copies of data and one will be sent to prod array and another copy of data will be sent to RPA in RP cluster at the prod site and then through WAN link, data will be transferred to the remote journal volume
But if the WAN link between 2 RP cluster goes down then the IO will be sent to the prod array and what will happen to the copy of IO that is sent to the RPA in primary RP cluster if the secondary RP cluster is not reachable because of WAN link down ? Then once the WAN link comes up what will be the process of synching the secondary RP cluster with the primary RP cluster ?
The preferred RPA would mark the location of the changed data as dirty in the bitmap stored in the production journal (only the location of the changed block is written, NOT the actual data). This is also called marking mode. Once the WAN link becomes up, the changed data would be read from the production volumes by the appropriate locations as marked in the production journal.
Hope that helps,
RecoverPoint Corporate Systems Engineering
Thanks for your response.
What will be the process if there is an extended WAN link outage ?
We have a scenario were network team is planing to upgrade the WAN switch code on both the switches simultaneously, expected time to upgrade/reboot is 5 mts to 15mts.
We would like to know how long the marking mode would continue ?
When there's a WAN outage / failure in communication between the splitter and the RPAs then we go into marking mode where we're marking changes in the production journal, these are only for bitmap purposes which means that we're writing the offsets to blocks which haven't been replicated yet. When communication is restored then we read the data from the production volumes with the locations written in the prod journal.
Hope that helps,
Product Engineer - RecoverPoint and RecoverPoint for VMs