The great and the good of the cloud computing community gathered at the Rackspace DevOps Breakfast Panel Debate this week in London's glittering Soho district.
Attending this month's "discussion panel breakdown session" were speakers from DevOps Guys, Dataloop, Skelton Thatcher, Eagle Eye and, obviously, Rackspace.
Stephen Thair - co-founder of DevOps Guys said the he believes in DevOps automation as so many of his customers have problems with Continuous Delivery.
Chris Jackson - cloud technologist and head of Rackspace's DevOps Practice Area said that he recognising there is a lot of automation in DevOps and that his company (with its very up front 'Fanatical' support offering) recognises that it now needs to address the intersection of support with automation.
The learning, learnings
The difference between ITIL and DevOps is that ITIL has a huge amount of information to draw upon, but that DevOps could (if it is done properly) is proposing an alternative model that has a perhaps more practical implementation these days with more iterative feedback into the ongoing state of the project....
... and it is this, centrally, that makes DevOps a learning process.
DevOps is a commitment to learning and experimentation (more so than a straight Waterfall development methodology).
Rackspace's Jackson is a huge fan of the CALMS acronym:
- Measurement or Metrics and,
This discussion moved (as might be expected) onward to whether DevOps was a technical issue or a human cultural issue -- despite audience protestations that it must be one or the other, the majority of speakers agreed that DevOps is both a human and a technical issue.
Speakers here suggested that a good route into DevOps (as a new cultural approach) could be to apply it to a smaller application inside the total IT stack and use this as test bed to bring wider DevOps approaches into an organisation -- the challenge here will be finding an application that is "separated enough" from the rest of the IT stack... but it can be done.
"Organisations must be set up to enable software systems to evolve over time -- DevOps enables this. DevOps enables the flow of metrics-based intelligence from production back to development," said Matthew Skelton, co-founder and principal consultant at Skelton Thatcher Consulting Ltd.
"Successful DevOps adoptions address the interaction technology AND teams to build and operate software systems effectively," added Skelton.
Old DevOps is a waste of time
Other suggestions emanating from this event included the suggestion that 'traditional DevOps' (i.e. not delivered as a cloud service) could in fact be a (comparative) waste of time for what are skilled systems administrators...
... what do we mean by waste of time?
If a sysadmin has to spend HOURS of time working to build operational servers, then isn't that a waste of skilled time if that server could be bought from a cloud supplier? The sysadmin could be doing something else more complex, more business-value-add and more live.
Yes, you would expect cloud (an DevOps as a service) vendors to say this kind of thing, but it is arguably quite an interesting proposition.