vSphere Metro Storage Cluster (vMSC) on NetApp clustered Data ONTAP - Overview.

By | August 3, 2015

Recently I discussed with my colleague the vSphere Metro Storage Cluster (vMSC) running on NetApp clustered Data ONTAP 8.3 or later as it is supported already by VMware. Our customer wanted to build a highly available virtual infrastructure stretched between two data centers and based on vSphere HA.

In this post you will find answers for the following questions:

  • What Is vSphere Metro Storage Cluster (vMSC)?
  • What is NetApp MetroCluster?
  • NetApp recommendations for VMware HA.
  • Failure scenarios.

vSphere Metro Storage Cluster (vMSC) overview

vSphere Metro Storage Cluster (vMSC) is a certified (you should always check/verify compatibility on VMware Storage Compatibility Guide) configuration for stretched storage (synchronized replication) cluster architectures. It means that you can create a vSphere cluster with hosts in two physical data centers (e.g. Tokyo and Yokohama). vSphere Metro Storage Cluster vMSC on NetApp clustered Data ONTAP 2Two vMSC configurations are possible based on how the hosts are connected to the storage arrays.

  • vMSC Uniform - In this configuration each host can access LUNs exported from both the sites. It helps in protecting against a storage failure at a site.
  • vMSC non-Uniform - In this configuration each host can access storage resources available only on its local site.

NetApp Metro Storage Cluster represents vMSC Uniform! 😉 For example EMC VPLEX can be configured as Uniform or non-Uniform.

NetApp MetroCluster overview

NetApp MetroCluster is a solution (based on synchronous replication, RPO=0) providing automatic recovery for any single storage component failure.vSphere Metro Storage Cluster vMSC on NetApp clustered Data ONTAP

As shown on the above figure, there are two NetApp clusters with two controllers (nodes) in each cluster. Each node has an HA partner in the same local cluster and a DR partner in the remote cluster. A1 and A2 are HA partners, as are B1 and B2. Node A1 and B1 are DR partners, as are A2 and B2. In an HA failover, one of the nodes in the HA pair temporarily takes over the storage and services of its HA partner.

The MetroCluster itself does not detect and initiate a switchover after site failure. It can be done manually (e.g. by you 😉 ) or using software - Tiebreaker. Tiebreaker is a program installed at a third site with independent connections to each of the two clusters. Tiebreaker monitors and detects individual site failure and intersite link (ISL) failure.

vSphere Metro Storage Cluster vMSC on NetApp clustered Data ONTAP - high level

Source: NetApp TR4128

NetApp recommendations for vMSC/HA

There are some NetApp recommendations for vSphere 6.0 HA:

Host Failure Detection:

  • Specifying a minimum of two additional isolation addresses and that each of
    these addresses be site local.

Host Isolation Response:

  • The Host Isolation Response setting should be Disabled.

VM Component Protection:

  • The Response for Datastore with PDL setting should be Power Off and Restart VMs.
  • The Response for Datastore with APD setting should be Power Off and Restart VMs.
  • The Response for APD Recovery After APD Timeout setting should be Disabled.

For more information please follow two NetApp documents:

Failures scenarios

There are some failure scenarios (more you can find in the mentioned guides):

ScenarioNetApp Clustered Data ONTAP behaviorVMware HA Behavior
Single Storage Path FailureNo change in MetroCluster behavior - the datastores continue to be intact from their respective sites.None
Single ESXi Host FailureNo change in MetroCluster behavior - the datastores continue to be intact from their respective sites.Restart VMs in the local site.
ESXi Host IsolationNo change in MetroCluster behavior - The datastores continue to be intact from their respective sites.Depending on the isolation response
configured, the host can choose to power off, shut down the virtual machines, or even leave the virtual machines powered on.
Disk Shelf FailureDuring this period, there is no impact on the virtual machine I/O operations, but there is degraded performance
because the data is being accessed from the remote disk shelf through ISL links.
None
Single Storage Controller FailureThe local HA performs failover.None
Complete Site FailureThe storage administrator or tiebreaker performs a switchover.Virtual machines running at the failed site fail. VMware HA automatically restarts them on the surviving site.

Conclusion

MetroCluster provides three basic methods for data continuity in the event of planned or unplanned events:

  • Redundant components for protection against single component failure
  • Local HA takeover for events affecting a single controller (not available in 7-Mode)
  • Complete site switchover

At last, there is a possibility to build a highly available VMware infrastructure based on NetApp MetroCluster with two nodes in each site!

3 thoughts on “vSphere Metro Storage Cluster (vMSC) on NetApp clustered Data ONTAP - Overview.

  1. Alex

    Hello,

    I always see the example in vSphere Metro Storage Cluster with datacenter A and datacenter B (as DR site), but this solution is possible if we want two datacenters active-active (DC A and DC B) and also a third site as the Disaster Recovery active?

    If it is possible how it would be?

    Thanks a lot

    Reply
    1. Mariusz Post author

      Hello,

      I haven't seen vMSC between three sites (active). I don't know any storage vendor supporting such configuration. However you can build three active sites using: vMSC (two sites) and third site with replication solution (e.g. RecoverPoint) as a DRC.

      Reply
  2. L

    Hi there, nice article.
    If I have a VM using root disk (vmdk) inside a VMFS datastore and several RDM LUN attached and NFS. VM can be restarted in any host as well as primary and secondary datacenter? Using netapp metrocluster.

    Thanks a lot.

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *