Not Dead Head

Side A

Category Archives: Agile Ops

about the “Introducing Kanban in operations” by Mattias Skarin at DevOps Days ’09


At this point the one agile method I really like.

Through the whole sprint idea from Scrum away and I will like Scrum even more. Besides the whole ridiculous estimation Poker thing, which is probably only there to get numbers for the velocity an so on, the one thing I really don’t like, are Sprints because IMHO they reduce the agility of the team. The team only needs a well prioritized Backlog in order to work effectively and efficiently. Uhm!? WAIT! if you do that, then you have … Kanban! — what a coincidence!

On the other hand, Sprints and estimating the effort to get things done can help the team get a vague idea of the project’s time line. If vague is enough (you’ll get any better than that, any way), than you can implement them. But I have seen already to many “broken sprints” to really support that.

Mattias also has also a Blog with a lot of information about Kanban ( There you can find his slides to the presentation (


Doing Ops big time

Jetzt geht es darum, wie wird Ops bei manchen Großen gemacht. Bitte nimmt euch die Zeit, holt euch Pop Corn, Cola und eine Decke und schaut euch diese Videos mit euren liebsten an, die/der könnte nebenbei z.B. häckeln.

§ Ops@Facebook

  • Versioning
  • automation, automation, automation
  • use a configuration management system (keep it agile! – the process)
  • EVERYTHING WILL FAIL, “und das ist gut so” , so be prepared
  • Messure/Instrument a lot, if not everything

§ Ops@Twitter

dude 1: Whatz up! Dude! I just came from our own brand new datacenter.

dude 2: What!? Dude, your company builds its own datacenters! Wow! That’s totally cool, dude!

dude 1: Yeah dude, and we have all this racks and cables and ….

That other dude sitting at another table: jejejeje!!! What a waste! Weil they were doing that, I was making money.

Sing along!!

Real Big Ops manage their own datacenters! Yeah yeah yeah

The even build their own datacenters! Wow wow wow

That is just so agile and cool and hey!

you even get to choose the colour of you cables!  … *LOL*

Ok, if you really need your own datacenter, then you are delivering an “über” global service.

„With NTTA, we can focus our resources on developing software and serving our valued user community rather than on building, financing, and managing an in- house IT infrastructure team.“  Lee Mighdoll, Vice President of Engineering and Operation, Twitter

… and the author isn’t getting any money (at this moment) from NTTA.

“Not even the best OpsTeam in the world (including everything outside USA) using the best tools they get/implement, can warranty availability and efficiency, if they must run crappy applications … it just won’t work.”  The author, by himself.

§ Ops_On_ChangeManagement

pure ITIL ~ high Process compliance ~ standardisation ~languorous and slow reaction times ~ sluggish                          |

|-à balance ~ Enlightenment ~ Nirvana

No Process compliance ~ low level standardisation (if any) ~quick implementation and fast reaction times ~ mayhem      |

Nirvana is the supreme state free from suffering and individual existence. It is a state Buddhists refer to as “Enlightenment“. It is the ultimate goal of all Buddhists. The attainment of nirvana breaks the otherwise endless rebirth cycle of reincarnation. Buddhists also consider nirvana as freedom from all worldly concerns such as greed, hate, and ignorance. No one can describe in words what nirvana is. It can only be experienced directly.”

Die Einzigen, die in Produktion testen, sind die Pharma Industrie. – read all presentation pages footers.

The 6 laws of reliability (by Joe Armstrong @Ericsson not Green Day).

A high available fault tolerant system must obey the following rules:

  1. Isolation
  2. Concurrency
  3. Failure detection
  4. Failure identification
  5. Stable storage
  6. Live Code Upgrade

Andrew: Who rolls back?

Guys: Hands up … yeah we roll back! 😉

Andrew: Screw that, roll forward.

Guys: Why?

Andrew: Because I roll forward, dog!

If you are agile (~fast!) enough to do CLD, then you’ll never need to roll back.

On DevOps and agile system administration.

Scrum for Projekt Management and Software Development and DevOps für …Operations?


Two points I’ll like to set fixed before I go forth:

  • Operations stays for availability and efficiency.
  • DevOps stays for “Push responsability to the edges.”


Adam Jacob on DevOps@ O’Reilly’s Velocity 2010 (~50 Minutes)




DevOps days – SOLD OUT! and I’ll not be able to attend.


Infrastructure Automation with Chef @ O’Reilly’s Velocity2010

Welcome to This is your first post. Edit or delete it and start blogging!


%d bloggers like this: