Many of my readers will already know that vCenter heartbeat is being removed from VMware’s product line up.   They will officially end support in 2017 but the kill notice has already gone away.   Today I attended a VMUG user conference and one of the speakers was Justin King (read more about him here) He joined VMware when they acquired the company that developed heartbeat.   So he was the perfect person to ask the golden question.  Now that vCenter Heart beat is gone what should I use to protect vCenter?

Background:

In the ESX 3.xx days it was common to find vCenter running on physical hardware.   This was due to multiple dependencies for HA to operate on vCenter.  It was also about customer confidence.  These issues have been slowly removed.  With the rewrite of FDM in 5.0 the recommendation has become to use a virtual vCenter and even more the Linux based appliance.   Heartbeat provided a hot standby for vCenter in the event of failure but it really took about 30 seconds to take over.   At this point HA takes 15 seconds + reboot and service start to restart vCenter.

 

Justin’s Take:

Justin provided a really great view that I wanted to share:  Most companies don’t have SLA’s on vCenter.  Most of the availability features work fine without vCenter.   So for most companies the 15 seconds + reboot is not a big deal.   He also suggested that HA Application should be used to restart applications as needed.   He also suggested that a management cluster is becoming more and more common.   Certainly with NSX and Auto deploy a management cluster becomes critical.    He also eluded to some new features that may solve the issue.

 

My Take:

This problem reminds me of the 2TB VMDK limit.  Everyone needed more space and it felt like VMware took about two years too long to solve.  For years customers have complained that everything about VMware is redundant except vCenter… I have to agree with them.   Talking to Justin did bring up one critical thought: customers complaints and needs are two different things.  Very few customers have SLA’s around vCenter… lots of SLA’s around virtual machine work loads.   So they have focused they energy on customer needs instead of wants.   Will they go to a redundant vCenter solution?  Yes they will just not right away.   There are two environments that require vCenter and have SLA’s  EUC and Cloud.   In both situations your customers will be unable to consume services without vCenter.   I think are both gain more traction the business case for redundant view will increase.   I think the move toward the appliance also opens up doors to solve this redundancy issue.

© 2014, Joseph Griffiths. All rights reserved.

Leave a Reply