In the past year, I’ve written a lot about how the DevOps (development-operations) movement assists companies like Twitter, Salesforce and others in continuously delivering software to their customers. I’ve also seen how version management companies such as Perforce (for whom I’ve written a few white papers) are cashing in on the continuous delivery trend to emphasize the importance of having a strong configuration management strategy at the hub of this high-speed world.
DevOps, in a nutshell, takes the concerns of the “last mile” of software delivery (normally handled by sysadmins or operations staff) and brings them forward to the first stages of software design (typically involving software architects anduser experience experts).
When I came across DevOps Borat a few months ago on Twitter, I began following him assiduously. Finally, I got up the nerve to interview this iconoclastic Kazakhstani geek, whose Twitter bio reads: “Cultural Learnings of DevOps for Make Benefit Glorious Teams of Devs and Ops.” At last, in just 10 succinct questions and answers, I give you …everything you need to know about DevOps. My work is done.
1) I thought DevOps originated in Northern Europe. Is Kazakhstan even farther ahead of the curve?
Is true DevOps is origin in Northern Europe in paleolithic time when man first apply principle of “not invented here” and discover square wheel. In Kazakhstan DevOps is as advance as of being indistinguish of magic. Is why person in charge of DevOps is call shaman.
2) How many DevOps success stories have you heard?
Every consecutive 10 minute when you not get Nagios alert is DevOps success story.
3) What is the best metric to watch when it comes to DevOps?
Best metric is number of new metric dashboard create per day. Winner is Etsy.
4) What’s the biggest DevOps technology “gotcha” to watch out for?
Are only 3 gotcha in DevOps: 1) disk is 100% full ; 2) database is in memory ; 3) you not receive Nagios alert for 4 consecutive hour.
5) How do you differentiate continuous integration, build automation and continuous delivery?
Is progression of build automation -> continuous integration -> continuous delivery which is also progression of developer is annoy -> tester is annoy -> ops is annoy because site is down.
6) How do you ensure deployment information gets back to dev and test?
Is very simple, send deployment error to pager and make dev and test of carry pager. As of deployment strategy, in startup we are allow every DevOps for fail deploy and bring site down maximum of 3 time. After 4 time they are get automatic promotion which is ensure they can not able do more damage.
7) What should you look for in a deployment package repository?
You are lost me at deployment package. In DevOps we are only deploy by ./configure; make; make install.
8) What’s the role of open source tools like Git, Gerritt, Jenkins, Maven, Puppet and Chef in the DevOps stack?
Main role is for put on resume when recruiter are play buzzword bingo. Was joke, serious answer is in DevOps tool is not matter because is about culture. Is why most DevOps are use 750 line shell script for every thing.
9) What do you think of nightly builds?
Is good idea but only if is combine with nightly deploy. Nobody is care if site is down at night.
10) What’s your current preferred DevOps tool chain (for check in, build, test, deploy)?
You are lost me at tool chain. I am only have 1 tool in DevOps arsenal: Perl.