Fight the FUD: vCenter on VDS on Nutanix NFS Datastore – Not a problem!

I saw this tweet (below) and was inspired to write this post as is appears there is still a clear misunderstanding of how the VMware Virtual Distributed Switch (VDS) functions when vCenter is down.

tweetmanish

My interpretation was the tweet was suggesting/implying the following:

1. If vCenter (VC) is on a VDS there is a problem in the event of an outage

2. Having vCenter (VC) running on an NFS datastore is a problem

3. Nutanix environments have problems with VDS deployments

4. In the event of an outage where vCenter (VC) is on a VDS and the underlying storage is presented via NFS by Nutanix, that this is somehow worse than if the storage was presented by another storage vendor.

Long story short, none of the above are problems and the author of the tweet is simply mistaken.

I highly recommend watching this recording of a VMworld session by @chriswahl (VCDX#104) & @thejasonnash (VCDX#49) which covers Distributed Switches in depth.

NET2745 – vSphere Distributed Switch: Technical Deep Dive

Here is a Video showing how a Nutanix environment recovers with vCenter offline with everything including the Nutanix CVMs connected to a VDS.

In the video, the Nutanix controller VM is using a dvPortGroup with Ephemeral Binding, however Static Binding is also fully supported.

So we don’t need to imagine an outage, the above shows the process start to finish and its only a few minutes to be fully operational!

No FUD!

bullshitrefute

Related Articles:

1. Example Architectural Decision – Port Binding Setting for a dvPortGroup
2. Distributed vSwitches and vCenter outage, what’s the deal?@duncanyb (VCDX #007)

Example Architectural Decision – Site Recovery Manager Deployment Location

Problem Statement

To ensure Production vSphere environment/s can meet/exceed the required RTOs in the event of a declared site failure and easily perform scheduled DR testing, VMware Site Recovery Manager will be used to automated the failover to the secondary site.

What is the most suitable way to deploy Site Recovery Manager to ensure the environment can be maintained with minimal risk/complexity?

Requirements

1. Meet/Exceed RTO requirements
2. Ensure solution is fully supported

Assumptions

1. vCenter is considered a Tier 1 application
2. vSphere 5.1
3. SRM 5.1
4. A single Windows instance hosts vCenter, SSO and Inventory services and is protected by vCenter Heartbeat

Constraints

1. SRM is not protected by vCenter Heartbeat

Motivation

1. Reduce the complexity for BAU maintenance

Architectural Decision

Install Site Recovery Manager on a dedicated Windows 2008 instance

Justification

1. When installing / upgrading /  patching  SRM including Storage Replication Adapters (SRAs) this may require a reboot or troubleshooting which may impact the production vCenter, including SSO and inventory services.

2. Having SRM separate to vCenter ensures the fail over is not unnecessarily delayed in the event of a disaster due to contention with vCenter on the same VM

3. SRM and vCenter work together in the event of an outage, as such they are less complimentary workloads

4. If hosted on vCenter, SRM will then be subject to the same change windows and be impacted during any maintenance performed for applications running on the same OS instance

5. The SRM application has different availability requirements than vCenter, as such if SRM was combined with vCenter, SRM (having a lower availability requirement than vCenter) would have to be treated with the same change management / care as vCenter which would complicate BAU maintenance

6. The SRM service (business) has different maintenance requirements to vCenter, as such they are not suited to be placed on the same VM

7. Having SRM on a dedicated VM aligns with the scaling out recommendation for virtual workloads

8. Having additional components on the same OS increases complexity and may reduce the availability of vCenter

Alternatives

1. Place SRM on the vCenter server

Implications

1. One (1) additional Windows 2008 R2 licenses will be required

2. One (1) additional Windows instance will need to be maintained in BAU

I would like to Thank James Wirth VCDX#83 (@jimmywally81) for his contribution to this example architectural decision.

Related Articles

1. VMware Site Recovery Manager, Physical or Virtual machine?

2. Swap file location for SRM protected VMs

CloudXClogo

 

 

vForum 2012 Sydney – TechTalk – VMware View 5.1 Desktop Deployment Solutions

At this years vForum in Sydney, I presented a TechTalk on VMware View 5.1 Desktop Deployment Solutions.

Below is the link to the recording, appologies the audio is not great as the TechTalk was done in the vForum lounge.

Josh Odgers – vForum 2012 TechTalk – VMware View 5.1 Desktop Deployment Solutions

As the video is not very clear I have provided a copy of the presentation below.

vForum VMware View Provisioning Options V0.