Revolution sounds a bit dramatic, doesn’t it? The truth is that adopting DevOps practice within our hosting company has been nothing less than that. We’ve had to tear up and start again with regard to many of our long-established practices.
Since we started hosting websites in 1996, everything we have done has been geared towards solidarity and reliability. That has traditionally been achieved through ring-fencing our entire operation, keeping close control over any changes and carefully managing the pace of any changes that do happen. In other words, the exact opposite to the agile world of DevOps.
Change often breeds fear, and nobody fears change quite as much as web hosts, who crave stability and predictability. But having got past the fear and gone through the revolution, we can confirm that it was all worthwhile.
We cannot claim to have had a great epiphany moment when we knew DevOps was the way forward for us. We were dragged kicking and screaming to where we are now by clients - but we’re glad we’re here. Here are some of the lessons we learned along the way.
For the reason’s outlined above, a DevOps approach is always going to face initial resistance from a web host who isn’t accustomed to that way of working. That is going to be hugely frustrating for a client who is raring to become more agile.
Speaking from experience, a face-to-face meeting is the way to resolve the issues and for both parties to understand the other’s position. Sorry, hosts, the upshot will be a greater understanding that your old way of working is no longer fit for purpose.
Move the ring-fence
If we accept that good, secure hosting work is achieved by penning in the entire infrastructure, good, secure DevOps work can only be achieved by moving the boundaries and bringing the development environment inside the ring-fence.
Failing that, you will have to find a way of allowing safe, speedy passage through the ring-fence for your clients.
IM is the way forward
Web hosting support traditionally revolves around a service request portal or ticketing system, followed by back-and-forth correspondence within the portal or by email. That doesn’t cut it with a DevOps approach.
Communication needs to become as agile and collaborative as DevOps practice itself. We’ve adopted instant messaging tools, such as Slack, in order to make ourselves - and by that we mean a named support contact - available immediately to the client.
Become part of the team
When you’re the external host for a client that is utilising DevOps, you have essentially become the ‘Ops’ part of the offering. As such, it is impossible to continue to operate as an external service provider. Instead, you have to become a remote part of your client’s development team.
Adopt like-for-like workflow
We soon learned that in order to become an effective part of the team, we needed to match up with our clients’ workflow processes.
For those working in enterprise Ops, development, staging, user acceptance testing, pre-production and production environments have been the same for a while. That isn’t necessarily the case for external hosts, but it soon will be.
A host’s development style and technical stack need to mirror the client’s approach.
Embrace real-time application performance
The quality of feedback and performance metrics made available to clients by web hosts varies widely from company to company. We’ve always prided ourselves on being both thorough and transparent in this regard.
But, once again, a client showed us that we weren’t doing enough to support a DevOps approach. We now use browser devkit outputs, plus New Relic and AppDynamics, to help our clients to see how their rapid development deployments are performing. As with communication, performance monitoring now needs to be as rapid and as agile as the DevOps process.
There you have it: some of the main points arising from our own DevOps revolution. We haven’t masterminded the changes - they have been driven by necessity and the requirements of our clients - but they have helped us to create an external hosting environment in which the DevOps philosophy can flourish. Going forward, it is difficult to imagine a sustainable future for web hosts who are not prepared to create that kind of environment.
About the author
Andrew Maybin is the Managing Director of web hosting and digital infrastructure firm Tibus. He is also a contributor to several websites and publications, providing insight on web hosting, the digital community, and wider technology issues.