This week we saw our first ‘at scale’ public production use case at Cloud66.  Based in Austin TX, and London, this is a brilliant young company that puts customers first and is showing the way for the next generation of automated hosting.

In this post I want to tell you all about it, and what it means for customers using Weave. Having launched Weave last September we are already on quite a journey – with much more to come.

Cloud66 – the next generation of managed hosting

Khash Sajadi, CEO of Cloud66, writes: “The Cloud 66 platform manages thousands of servers across 76 data centres in 8 cloud providers for hundreds of small and large companies. Thousands of developers rely on Cloud 66 every day to build, deploy and manage their applications on their servers.

This is exciting because Cloud66 achieve this with almost no human intervention. Their platform is almost completely automated, and they run super lean.  I am not allowed to share numbers but the ratios are very impressive.

For customers, Cloud66 combine three value propositions:

  • Pay-per-server pricing at similar levels to some of the well known developer friendly full-stack solutions.
  • Zero lock in: choose you own application architecture, choose where it runs.  This includes your own data centres or one of seven pre-integrated clouds.
  • The product is simple enough that anyone can use it, and no big services engagement is needed.

What I like most is that Cloud66 have achieved the level of efficiency that enterprise customers keep hearing about and want.  This lets customers manage costs down to ‘per team scale’ for as long as they can, instead of having to make a big company-wide purchase upfront.

Docker and the problem of automation at scale

Cloud66 have recently expanded their solution and now use Docker for software portability.  This introduces simple and homogenous deployment and management across multiple application architectures and developer tools.  But the missing piece is that most applications are clusters of multiple components.  Khash tells me that many of their customers are running 20 to 100 nodes per application.

So: how do Cloud66 provide that across all use cases with a single solution?  Recall that Cloud66 are currently supporting multiple data centres (76 and growing) and their customers are thousands of developers each with their own needs across test/dev, UAT, production and D/R.

Weave – radically simpler applications anywhere

Weave solved the portability-at-scale problem for Cloud66, so that they can support all their customers with just one technology.

The backbone of Cloud66 is a component called ContainerNet.  In Khash’s words, this “is backed by Weave which brings a native-like network interface to the containers requiring no code changes, is highly performant and secure and scales very well.”.  First let me say thank-you to Khash and the Cloud66 team for writing about their use of Weave!

Now – let’s drill into the use case and understand what this shows:

  1. “No code changes” – this is essential so that the solution is easy to use out of the box by every Cloud66 customer, no matter what tools they want to use, and whether they have a traditional application stack or some hip new architecture.
  2. “Performant and secure” – Cloud66 built ContainerNet quickly and without lots of Q&A.  No fiddly tuning required.  No discussions about physical topologies and constraints.  No professional services help required.
  3. “Scales very well” – Cloud66 can support as many data centres and customers as they need to, and to date are seeing good results with customers whose applications are the 20-100 node range.

In summary: Weave works the same way as Docker – it is portable, simple and idiomatic to developers.  This makes it possible to focus on building your application instead of getting bogged down in infratstructure wrangling.

What’s next?

If you are a company wrestling with how to get the most out of containers – please contact us. Cloud66 is a great example of what can be done, by taking common sense approaches that provide what is needed and no more.  With Weave itself, we are heads down pushing towards our 1.0 release which will provide enhancements to how container clusters and networks can be managed.  These include lightweight service discovery and automated address allocation.  Astute users have already been using these, e.g. weavedns, but with 1.0 we are moving them to a first class state.

Finally – a shout out to anyone working on containerization in or near London or SF.  If you want to come to (or speak at) one of our events, or meet any of the team f2f, please get in touch 🙂

— alexis

UPDATE — Listen to an interview with Cloud66 and Weaveworks about this case on the Cloudcast – “Container Centric Application Deployments”