vSAN1

 View Only

  • 1.  Synchronous data replication

    Posted Aug 30, 2017 04:12 PM

    We have 8 nodes running vSAN hybrid configuration in our main datacenter. We replicate ~20 critical VMs to our DR site with Veeam once a day. The replication job takes ~5 hours to complete over a 400Mbs connection. The critical VMs consists of DC, Exchange, web, app and database servers. There is a plan to build a new emergency center/server room on-campus. The goal is to have synchronous data replication of those critical VMs to this new site (still keeping the asynchronous replication to DR).

    • Do you think vSAN stretched cluster will be good for this solution or any other suggestions?
    • I remember reading a VMware document, the preferred site and the 2nd site for the stretched cluster need to have the same number of hosts. However, I also read somewhere that it’s not necessary and it’s support by VMware. Since we only have to replicate the critical VMs and not all the VMs, I am hoping we can just get 3 or 4 nodes on the 2nd site. Can someone please clarify on this?

    Thank you!



  • 2.  RE: Synchronous data replication

    Broadcom Employee
    Posted Sep 04, 2017 08:34 AM

    Hi,

    Yes, a vSAN stretched cluster should be a good fit but you will need to mirror the amount of hosts and disk groups on the emergency site.

    More details here:

    VMware® Virtual SAN 6.1 Stretched Cluster & 2 Node Guide

    Page 14 - Configuration Minimums and Maximums - Hosts per cluster

    https://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/products/vsan/vmware-virtual-san-6.1-stretched-cluster-guide.pdf

    This 31 host (15+15+1) max limit for stretched clusters remains in 6.5.



  • 3.  RE: Synchronous data replication

    Posted Sep 05, 2017 06:53 PM

    Hello pineapplehead,

    Yes, 'Unbalanced' or 'Asymmetric' stretched-clusters are supported (in vSAN 6.6) - but of course this needs to be configured properly and within reason.

    This is achieved via what is essentially data locality settings via Storage Policies - PFTT (Primary Failures To Tolerate) and SFTT (Secondary Failures To Tolerate).

    Some good resources on this:

    yellow-bricks.com/2017/07/30/unbalanced-stretched-vsan-cluster/

    blogs.vmware.com/virtualblocks/2017/04/18/pm-hub-better-site-protection-stretched-clusters/

    storagehub.vmware.com/#!/vmware-vsan/vsan-stretched-cluster-2-node-guide/configure-stretched-cluster-site-affinity-1

    Bob