#1: What does good DevOps look like?
Good DevOps is all about simple measures that IT and business can use together. Processes such as mean Time To Detect (TTD) to provide metrics on an incident and measure its frequency, size and impact can make a big impact. Following directly on from there, we look at mean Time To make a Change (TTC) and start to look at repetition rates to see if the same problem keeps happening.
Any Enterprise can use a use a simple service management system to collate, track and report. Only then can you start to benchmark and then see a tangible and measurable DevOps service improvement over time. But don’t forget the human element! A simple key to successful DevOps teams is retention of key permanent talent and staff. High turnover leads to high failure rate of the DevOps process and, subsequently, the wider project.
#2: Do we need to work towards a common language to unite Dev and Ops in order to make DevOps work?
I was co-founder of Fox IT, one of the companies that developed and authored ITIL into the commercial world. Over 90% of global corporates even down to SMEs have implemented a service management framework like ITIL in their IT departments covering IT operations, so this should show you the importance of codified terminology and language.
The key is to understand how developers can interface into this world. Ops sometimes see Dev as the Wild West, however ITIL needs to lean out, hook into and in some ways optimise and leverage the DevOps principles of culture, lean, automation, measurement and sharing. ITIL is not moving fast enough and IT is still struggling! DevOps is just starting to catch the attention of the ITIL base – and the thousands of Ops departments who speak ITIL. Find the bridge and hook into the ITIL world they will love and embrace Dev forever more.
#3: How do we ensure the next generation arrive DevOps ready? What skills need to be developed?
The demand for DevOps skills is growing; it increased by 75% from Jan 2012 to Jan 2013. DevOps as a skill has increased on LinkedIn by 50% in the same period. As with ITIL, an open set of common training IP needs to be produced by the community for consumption by new talent.
We believe we can assist by placing junior engineers into the IT industry and training them over a six-month period whilst on secondment. After six months they have the skills to become a DevOps engineer.
The three disciplines we believe should be trained are:
- People & communication skills
- Process re-engineering — an understanding of how to examine and re-engineer business and IT processes are essential skills for DevOps
- Experience with automation & deployment tool sets.
#4: Does DevOps need to be sold to the business, or should we just make the magic happen? Does the business need to change to make DevOps work?
DevOps needs to be sold to both Dev and Ops. The business doesn’t care; all they want is lightning fast and business changing deployments. The focus is with the IT folk entrenched in the silos. Pick a project that can be started and finished using a new methodology. Set the team goals as a pilot and set up a journal of the activity, issues and solutions.
#5: What are the biggest barriers to starting DevOps?
Probably disbelief from those that would say: “Oh it’s another service management framework” etc. Other big barriers include troublesome non-ratified standards (where they exist), IT silos (and the need to break them down) and no stakeholder buy-in from the CIO. Have a look at the DevOps readiness assessment from Continuo.