UNSOLVED

ca_fahem

updated

4 years ago

C

ca_fahem

1 Rookie

•

4 Posts

0

2303

November 7th, 2022 15:00

PowerStore Metro Volume Architecture

Hello, 

I have to site with vSphere clusters and two vcenters in linked mode configuration, we want to use two Powerstore clusters to replicate between the sites using Metro volumes: 

  1. Is vSphere stretched cluster (vSphere Metro Storage Cluster configuration) a requirement to use Powerstore metro volume between two sites?
  2. Does Metro volume support only vSphere hosts? 
  3. Can i configure a metro volume with only one vCenter and one vSphere cluster in site A to replicate only the storge to site b without using the vMSC configuration? 

Thank you.

  • SteveK821

    1 Rookie

    •

    30 Posts

    974

    0

    Posted November 8th, 2022 05:00

    1) For Metro Volumes, you need one vCenter and a stretched cluster.

    2)Only ESXi Hosts are currently supported with native Metro Volumes in Powerstore.

    All hosts should see both Powerstores (Uniform Connectivity), but you can also configure a host to only use one side (Non-Uniform Connectivity)

    https://www.dell.com/support/manuals/en-us/powerstore-3000t/pwrstr-protect-data/configure-host-connectivity?guid=guid-97f7bba6-37c5-4fab-bae4-4296a3a5fc17&lang=en-us

    For question 3, yes you could use metro to replicate the storage to site B, without any hosts connected to Powerstore B (I think).

  • ca_fahem

    1 Rookie

    •

    4 Posts

    973

    0

    Posted November 8th, 2022 06:00

    Thank you for your answers

    For the question 3, if I replicate only the storage to site B , can i map it to  hosts in site B in other clusters than site A (so another vCenters and ESXis) (manually without HA & DRS).  does the PowerStore permit that ?.

  • SteveK821

    1 Rookie

    •

    30 Posts

    911

    0

    Posted November 17th, 2022 06:00

    I don't know if that would work. I'm sure Powerstore would permit that, but I think that would lead to data corruption. The Datastore is read/write on both sides and all writes are replicated. You'd have 2 different clusters writing to the same VMFS, that don't know about each other. Bad things will probably happen there. I don't know if ESX has any sort of failsafe mechanism to prevent something like that.