I’ve been at the OpenStack Summit in Vancouver this week, and what has struck me is the difference between the AWS ecosystem and the OpenStack community – they are at times like chalk and cheese.

So what are the main differences?

Amazon has all but captured the public cloud space. It doesn’t have this space all to itself (Microsoft and Google are growing rapidly to challenge it), but it doesn’t really have any interest in the private cloud space (unless you have a budget the size of the CIA). Conversely many of the main players in OpenStack only really play in the private cloud space – HP for example let slip that it wasn’t really a player in public cloud, before rapidly withdrawing the comment and claiming that it was, when in reality it isn’t. So it is getting increasingly easy to picture a future in which a handful of players dominate public cloud with their own proprietary stacks, while OpenStack becomes the de facto standard for private cloud.

Then there are the two different ecosystems:

As we covered in our preview for this month’s cloud rankings AWS’s ultimate appeal isn’t necessarily its cost (even with continually falling prices, it can still be quite expensive) or its ease of procurement as an elastic hosting provider. Its main appeal is now its massive ecosystem of services and the ability to tap into them fairly quickly. As fast as AWS innovates, its ecosystem adds value faster than it could ever do on its own. It’s all relatively quick and relatively easy to implement and there’s lots of services to choose from.

The OpenStack ecosystem is very different: as we are seeing at the Vancouver OpenStack Summit, OpenStack boasts an enthusiastic following among the developer community and a very long line of powerful companies are supporting the project and contributing to the code and marketing of OpenStack.

I have also been impressed by the progress that the OpenStack community has recently made to become more organised and address some of its most obvious weaknesses. This week it announced interoperability testing requirements for products to be branded as “OpenStack Powered” including public clouds, hosted private clouds, distributions, and appliances (initially supported by 14 community members). A new federated identity feature will become available in the OpenStack Kilo release later this year (which is being support initially by over 30 companies), and it has announced a community app catalog. Add to this the work that it has done to certify major vendors via its DevCore programme and it is making all the right moves – but are they too little, too late? As I mentioned above it has already lost the public cloud to Amazon and others, and realistically only has the private cloud in its scope – albeit this is one fairly large niche.

The development ethos within each community is very different as well:

In the Amazon marketplace you have partners developing stand alone services that are designed to plug into one another. Typically they are closed source, well integrated via AWS APIs and normally as easy to spin up as any Amazon service. The OpenStack community much of the focus is on collaborating to develop new functions, rather than standalone services. The venture capital firms that were so keen to pour money into OpenStack startups a few years back are only now starting to realise that much of the development has been contributed to the common good in the form of open source code, and that the venture capital firms are unlikely ever to make a return on this.

OpenStack’s Achilles heel is its usability. Speakers wouldn’t be joking about the need for a PhD to implement OpenStack if there wasn’t at least a modicum of truth to the matter. For large clients with the tech skills to implement and maintain it, this isn’t a problem. And the SMB community that have minimal tech skills look to MSPs to provide cloud to them as a service. It is in the middle ground between these where OpenStack risks ceding ground to AWS and others – although innovative firms like Platform9 are seeking to address this.

And finally the differences between the business models:

One major OpenStack client that I spoke to framed this very well. He said that his company has many apps to support, each of which is used by a limited number of users. Amazon is set up as a volume operation where you need to have a service that appeals to many clients, whereas OpenStack caters more for the bespoke needs of each client. He went to Amazon for any off the shelf apps, but to OpenStack for everything that needed any tailoring, as Amazon would either be uneconomic or unsuitable for the number of users he had on each app.

Given all these differences you’d think that AWS and OpenStack were indeed like chalk and cheese. However the underlying technology for each is relatively similar – for every component in AWS there is an equivalent in OpenStack – this is all illustrated very well in a blog by RedHat listing the equivalent components in each stack, here: http://redhatstackblog.redhat.com/2015/05/13/public-vs-private-amazon-compared-to-openstack/

Suddenly the differences don’t look so great. Maybe they’re both cheese after all. AWS is the pre-packaged cheese that you get in the supermarket – well packaged, uniform and relatively cheap. OpenStack is the cheese that you get in at the deli counter – cut to size, often richer in flavour and arguably worth paying a little more for when you really want and can afford to.