WoodITWork.com

After attending a group discussion on vCenter I left perplexed by what was shared of the product roadmap and current customer feedback. On reflection I wondered whether something big enough was on the horizon that they were deliberately throwing us off the scent as I couldn’t believe how some questions raised in the discussion were not recognised as problems. Virtualisation management is so incredibly important, customer experience is key. Although great steps are being taken to address some functionality gaps, vCenter design-wise remains the same which isn’t good enough any more. I’ll go through the discussion and then add my thoughts.

This post adds some more colour to the highlights from my VMworld Europe 2015 coverage:

• VMworld Europe 2015: The Day –1 Buzz
• VMworld Europe 2015: The Day 0 Buzz
• VMworld Europe 2015: The Day 1 Buzz
• VMworld Europe 2015: The Day 2 Buzz
• VMworld Europe 2015: The Day 3 Buzz

Mohan Potheri and Madhup Gulati both from VMware led the discussion interactive by design rather than a presentation.

VCSA

They discussed migrating to the vCenter Appliance (VCSA), reiterating the maximums are the same for Windows and VCSA, There are currently export and import tools for the SQL data but it is still rather clunky. In the next version they are looking at having a much more seamless migration from Windows. VMware has Feature Walkthroughs to simulate what your upgrade will look like featurewalkthrough.vmware.com. VMware is going full steam ahead for the appliance and would like us all to join them.

Update Manager

There were concerns raised about Update Manager, you can still have Update Manager as a separate installation on Windows connected to a VCSA.

Good news is in 2016 Update Manager will be part of the VCSA, the UI integration is already done with vSphere 6.0 Update 1 and they’re working on porting the back-end code.

They are also looking at the possibility of putting SRM in the VCSA appliance.

vCenter is very resource sensitive in a good way which means it will actually improve performance with more resources

VMware is also working hard on removing the need for the Flash requirement in the Web Client, look at Host Client HTML5 fling to get an idea.

High Availability and Scale

Also they are putting in some sort of higher availability for the next version of the VCSA. It sounded like an active-passive model. They said they also messed up and canned Heartbeat too soon without an alternative. They should have waited a year or two until a better native solution was available. They said the reason for canning it was the adoption of Heartbeat was so poor and was bad.

I still find it incredible that for all the hard work they’ve been through to build a new vCenter with the VCSA they haven’t done some sort of scale out model. A golden opportunity lost when they preferred to port from Windows rather than be a little bolder.

Madhup said that 5-10 minutes downtime should be something you should tolerate. I disagreed, if you are running a cloud or VDI environment, 10 minutes at 9am when no-one can deploy a VM or log on to their desktops would be a serious problem. They say this is a very hard problem to solve, agreed, but VMware prides itself on being a distributed systems company and they won’t build a distributed vCenter. I wasn’t impressed that this wasn’t recognised. I asked about GemFire as an example. They said vROps uses Gemfire but as the actual system of record is held in vCenter if something goes wrong with vROps it can just pull the info back from vCenter as the source. If vCenter was the source then what? I don’t know but please do something.

PSC

Talk then moved on to the PSC and they went through some of the architectural details and deployment options.

Next release of VCSA will be able to auto failover to different PSC, no load-balancer required!. This should never have been designed to require a load-balancer in the first place in my opinion. DNS-like design should be used, each component should be able to specify a primary and secondary PSC as the PSC information isn’t continuously changing, think how simply Active Directory replicates with no load-balancer in sight

Then someone asked about the scale of PSCs. They mentioned they were running 17 Vblock racks which come with a vCenter per rack. Current vCenter PSC design is to not have more than 10 vCenters within a single PSC domain. The suggestion was to then split up your environment into multiple SSO domains, The questioner then mentioned that this would involve arbitrarily carving up their environment. I mentioned this would also mean you couldn’t vMotion across different vCenters in vSphere 6 if they were in different SSO domains. The presenters said this was a corner case at scale and wouldn’t impact many people. Not a very cloud scale answer was it?

I also suggested being able to have embedded PSCs between 2 vCenters replicate with each other which is currently not recommended is a must have. This is a very common deployment with 2 data centers with say SRM. The feedback was, yes, we should look into that. I found that astounding. There must be a disconnect with customers and the vCenter product team if this is actually what they think. Recommending (never mind the grey area of not recommended vs. not supported) that you need to deploy an external PSC if you only have 2 x vCenters for a small deployment is silly. It seems the overall multi-year plan to pull out SSO, move it to a PSC and have various components plug into it is far too complex and elongated which means customers are faced with a new deployment architecture every year. if they upgrade.

A few attendees fed back it would be great for vCenter to have a single operator view across on-premises + vCloud Air + vCloud Networks.

So, where to from here…

VMware has seriously dropped the ball with vCenter which was once VMware’s crown jewel. The success of virtualisation was directly related to the ease of management and vCenter enabled this. As virtualisation has grown and enabled cloud, vCenter is showing its age, Still using Flash, bloated functionality, no included Update Manager, no High Availability other than HA and a too slow restart, complicated external load-balancer requirements for the PSC, horrendous certificate requirements. They had to build EVO:RAIL as a layer over vCenter to simplify the use of their own product!

I’m optimistically hoping we were being fooled with this session and VMware is cooking up the most awesome HTML5 successor to vCenter but they can’t talk about it. Multi-cloud management and visibility, active/active highly available operations built in, transparently scale as you grow (think of those millions of containers to manage), trimmed down architecture and UI, move the stats/events/etc. into vROps if you need to, common APIs across vSphere/vCloud Air/vCloud Air Network. Have an option to have vCenter in the cloud, managing private resources with management and new functionality delivered as a service. Is this the future or am I dreaming?