vExpert and Ravello Labs

A month into 2016 and we’re rapidly approaching the announcement of the VMware vExperts for 2016. The benefits of being recognised by this programme have varied over the years (and as the numbers have increased, the ‘freebies’ have inevitably gradually dwindled), but one of the things I have found most useful is the free use of Ravello Labs for running nested ESXi. See here for more details about what is on offer.

Being able to spin up a personalised lab environment for testing has meant that I’ve not run a home lab nested on my laptop all year – in fact I’m not sure I could actually run one on it these days with the increased memory requirements and increasing number of VMs that seem to comprise a modern VMware lab. See here for a comparison of Ravello Systems labs against some physical home lab options

There are some limitations of course, the main one is that you can’t easily run the VCSA (there are ways, but it’s a pain to implement), and some other VMware linux based appliances seem to be affected in the same way. There is also a memory size limitation – I think this is 8Gb at the time of writing, but obviously you can add additional ESXi “servers” to your lab to scale out if necessary, and my laptop currently only has 8Gb total…

I still currently use VMware Hands On Labs to get hands on with specific products (NSX, VROPS, VRA), as they are quick to spin up and are generally pre-canned with the product I’m wanting to investigate, but for general VMware ESXi/VCenter work, I’m very impressed with Ravello Labs, and I’d like to say a big thank you to them for this very generous benefit, and please keep it running for 2016!




So cloud is becoming a commodity, where’s the real work and interest going to be?

I wrote a few months back about commoditisation of IT services and I’ve been thinking about this some more, wondering where my career is going to be headed over the next 10 years.

The likes of AWS, Azure, and Google, are continuing the “race to the bottom” as far as pricing goes, as well as enabling availability features. This is akin to “Stack it high, sell it cheap” and the profit will be a small margin, but on a huge amount of sales. An interesting analysis of the figures from AWS last quarter is available here (I can’t speak for the veracity of the analysis, but it’s an interesting read). It’s difficult to gauge how big a Cloud services business would need to be to compete in this world, but if multiple datacenters are needed around the globe, then I’d hazard that “pretty big” is a good answer.

What will these big players not want to do?

  1. Handhold inexperienced customers
  2. Work with a customer to migrate complex requirements into a finished solution
  3. Enable smooth interoperability with a competitor
  4. Advise on how you could improve efficiency

These are all areas that could become higher margin offerings for firms in the IT service industry. They should also be more interesting than standing up huge amounts of identical kit.

One thing we are going to need is a cloud-agnostic management suite, for deploying and managing cloud based apps. It will need to handle agile delivery, and be aware of multiple cloud services offerings and their pricing strategy, to best place those apps, and advise when to move them between cloud services, to get the best value. I don’t think there is anything currently out there that meets this bill.

There are a few options that may get there soon though, while it may have been in the news for all the wrong reasons, CSC’s Agility Platform (aka ServiceMesh) is heading in the right direction (disclaimer – I work for CSC), and VMware’s vRealize Suite is also covering a lot of the necessary ground. Both of these are still too complex at this moment in my personal view, but are evolving rapidly.

Maybe even 2 tiers of toolset are what is really required. A simple one, aimed at smaller, bespoke developers, and a more complex and scalable one aimed at large integrators. Perhaps some of the current tools are just about good enough right now, and in the spirit of Lean methodology, that Minimum Viable Product is all we need to get going.

Time will tell.

Commoditisation of IT Services

I’ve been thinking a lot lately about where things are heading with IT services. As an avid follower of Simon Wardley’s blog and his regular tweets, I believe it looks increasingly likely that the industry is headed for a big shake up.

I like analogies, even if they’re not perfect, it helps get my mind around hard concepts. The one I’m going to use here is the automotive industry.

When cars were first invented, they were custom made, very expensive, unreliable, and required constant maintenance. Henry Ford made one of the first large changes, with the industrialisation of the assembly. A standard product could be turned out much more cheaply than before, also more consistently which is also of vast importance.

Over the years, this has been the standard way of car production, mainstream products being stamped out in a consistent and uniform way. The immense costs of design and development generally limiting the number of models available, although recently with the use of car “platforms” such as VAG’s MQB platform that 8 models are currently built on, a greater range of models is now coming to market, to cover smaller and smaller niches. At the opposite end of the spectrum from commoditisation, small manufacturers still exist, but an ever increasing cost of design, and meeting safety and emissions standards, is making the cost per unit climb steeply.

And so it is with IT services. The assembly line is automation, such as cloud automation tools. If you aren’t heavily scripting your provisioning, you are a small manufacturer like Morgan – interesting, but destined to become an irrelevance to the market as a whole. Your cost per unit is going to be prohibitive.

As the cloud platforms mature, they make it possible to cover more use cases, people are standing up SAP systems on AWS, which would have been unthinkable even a couple of years ago.

What then of private clouds? You will still find them, but the massive cost differential with public clouds will make them very much a second choice. Pretty soon, they will be just where security concerns and governance dictate that they can’t go on a public cloud. I think it’s highly likely that these will also succumb eventually, G-Cloud is a prime example of this momentum.

So will AWS, Azure, and Google own all the business? Well, yes and no. Those 3 will probably have the majority of the hosting, however the majority of businesses will require an integrator to build the automation, to manage the service, to provide the value, rather than just a bunch of virtual servers.

There will be a lot of differentiation and a lot of opportunity in providing those cloud management services, and it will need engineers that are skilled in automation, in service management, and not just an OS or a Hypervisor.