The most awesome Mark Thiele recently posted a debate-provoking post about the direction that OpenStack might go in, title OpenStack – the future of Cloud & Infrastructure Management. My contribution to the debate, as posted as a comment on Mark’s blog, is that I don’t think it will happen (but it’s not all down to OpenStack’s challenges!).
To summarise I think this is a problem that is too hard for just one solution to dominate. There are too many competing factions, technology changes at a rapid pace, and that when you add this to OpenStack’s apparent trajectory towards a fractured ecosystem with HP, Redhat and others all tugging and pulling in their own directions… it just seems too difficult. I could be wrong!
I’ve been living with the “Cloud and Infrastructure Management Conundrum” for a long time now, both as a vendor of cloud technology and as a cloud service provider in my role as Canopy Cloud CTO. This is a tough problem to solve. Some, like CSC and Dell, attempt to solve this by buying great technology in this space (ServiceMesh and Enstratius respectively). Others are combining leading business, operations, infrastructure and application automation into a pluggable “cloud fabric” to solve the problem. Both approaches have challenges.
I tried to draw up an image that I have in my head, the problem space is in the middle with the UX/UI/API layer – if you have a rich portfolio of infrastructure, applications and services then it is unlikely one product (like OpenStack) can fill this role. If you have a large number of complex clients with different demands, this is also unlikely to be met by one product. As usual, the more variety the more complex the problem. IE – the more boxes and complexity around the middle box, then the harder it is for a simple solution to solve it. You can probably only, as a CSP, control what is in this middle box. You can choose to not add complexity (clients or technology) and that is a business decision.
And here’s the rub: complexity is INCREASING not decreasing. Customers are asking more and more of the cloud (don’t just run web apps, run my SAP!), and infrastructure tech is changing at rapid pace (SDx). Therefore, one should assume IT IS IMPOSSIBLE TO HAVE ONE PRODUCT TO RULE THEM ALL.
Anyhoo, it’s just my warped opinion but thought I’d write it down anyways. Tell me if I’m wrong! (It does happen from time to time).