A customer experiences a sudden loss in write performance on the HP 3PAR StoreServ array on the ficantly. What should you advise the customer to check primary site. Reads are not affected, but…

Changing from asynchronous periodic to synchronous replication would worsen write performance, as synchronous replication introduces latency from the secondary site into every primary write operation, exactly the opposite of the desired effect.

While Fibre Channel infrastructure issues can affect performance, if only writes are impacted and reads are fine, the issue is less likely to be a general Fibre Channel link problem and more likely related to replication synchronization.

Updating host multi-pathing software is a general maintenance task and unlikely to be the first check for a sudden, specific write performance degradation related to replication, especially when reads are unaffected.

If replication is set to synchronous, every write on the primary array must be successfully committed on the secondary array before the primary host receives an acknowledgment. A sudden issue on the WAN link or the secondary array's performance would directly impact primary write performance by introducing latency, while reads, which are local, would remain unaffected. Changing to asynchronous periodic replication would decouple primary writes from secondary acknowledgment, alleviating this bottleneck.