Where to start

As I spend more time engrossing into the world of DevOps there’s been a number of occasions where something has not felt right with the relationship between operations and development. Within the team at work we will be hiring developers to work on the integration back with the development team, improving the build process, re-factoring the code to make it quicker to code / build / deploy All of which is good stuff.

However we fore mostly run a service, that is the main reason for the existence of the group. Without the correct support framework we will not be offering a service but instead offering a really fancy technology exercise so we can say we do X or Y or Z and I worry that is the route we are destined for.

Can Developers do Sysadmin tasks?

Of course they can, why couldn’t they, in the same way a mechanic can paint a car, we are not in the business of stopping people achieve their full potential, so have a go Jo is welcome here. The bigger question is if they should be doing it, in much the same way as I am not the right person to code a large enterprise product, Developers are not the right people to be making decisions about about service restarts or process niceness.

So I believe that a Developer can do the tasks of a Sysadmin, I believe that with enough training they can get to a point where they are not making random changes to a system to fix a specific problem with out understanding the consequences. However, I also believe a good graduate would provide the same level of risk and knowledge to bring to the table, so having an understanding of programming is a plus but Sysadmins aren’t in the business or random changes.

Can a Sysadmin be a Developer?

Sure, Why not, Same role in reverse? Almost. So I can program or have programmed in a number of languages which I tend not to bring up so…

Pascal

Delphi

C/C++

Java

PHP

Javascript

Perl

Ruby (as of a couple of weeks ago)

Bash

Awk

So I have quite a few languages that within about 10 mins and a few nudges on google I can write something reasonable, I have made a lot of different applications (another list I hear you cry out for!)

Maze solvers

Text based adventure games

Arkanoid

Web based route planner, Granted I only drew the maps form 6 million data points with zoom functionality

Content management system…

Geoblog with google maps and email updates / geo tagged pics

Web shops

System monitor with averages and weekly summaries

bit stream cypher

cd to mp3 encoder with CDDB lookups

Just 10 things I’ve written, So I would say I know enough about programming, probably more than necessary for my role.
And yet I still have no interest in being a programmer. So shoe on the oppersite foot, maybe I should be doing some more developer focused work. It’s been a while but it could be just what I’m after.

Based on that I already am a part time developer much in the same way that a Developer is a part time sysadmin, I mean their programs run on systems right…

Who should do what?

Well, Developers code, sysadmins admin.. I don’t think it get’s harder than that. I think it is easy for everyone to agree that the Developer will be best spent writing code and helping out with specific system scripts or puppet manifests / capistrano. It is also very easy for everyone to say that the sysadmin should check the RAM utilisation, RAID configuration, Disk layout etc etc.

If all of that was correct this blog would end here, however; over the past few months something has been niggling at me and every now and then i’m involved in a conversation with an Developer which is ultimately “It’s not that hard just do X or Y” and it’s this which I have the biggest issue with developers on.

Let’s take rolling out a new Amazon AMI:

Developers approach

Deploy new server with AWS tools

Login

Done

Sysadmin approach

Start Deploy new server with AWS tools

Pause, because deploying keys isn’t a good idea or secure if everyone is using the same one…

Continue but with a generic “emergency key” configured

Check file system layout

Realise it’s all on one 6gb volume, fix issue

Create individual users

and so on…

Different skill sets, Any monkey in a suit can click on a few buttons in a web UI, Knowing that splitting out the linux file system to different partitions, or at least understanding the impact of that is important.

Summary

I think the two skill sets can work harmoniously, but there is still a boundary, caused by experiences and expertise and for DevOps it’s about using each other strengths and avoiding the weaknesses. There’s been times when i’ve been doing OOP PHP or working with inheritance or writing a very complicated script where having someone that I can say is this or that better than that or this? so having someone around would be good I imagine it works both ways, especially when it comes to configuring the system.

Developers tend to be very focused from my experience and because sysadmins are more generalists they are looked down on, I hope as DevOps becomes more common place and the realisation of harmony comes about. Lets see if in 6 months there’s another post on about how disastrous or successful this integrated approach becomes, it is new ground and it will be interesting to find out what happens if nothing else.