Date: 28th March 2013
It was a treat on Friday to be at the DevopsDays London Conference and to get the chance to speak to, and share the platform with, some of the leading exponents of the new joined-up DevOps world.
There was some fascinating discussion, with some tips and tricks in the “Monitor Everything” and the “Log Everything” sessions.
DevOpsDays started in Europe just two years ago. Events have now been held at a number of cities round the globe and this was the first in London. The events are effectively community led and have a model of a morning of key speakers, then the afternoon having an Open Space Technology approach to (un)conference. This was my first exposure to Open Space.
The mechanism was very effective in ensuring that, of the variety of participants who proposed themes on the day, it was those with the most votes that were held.
The DevOps concept itself is still quite young. It was nice to chat with some of the more senior guys like me who were there, sprinkled among the mostly young crowd, to get the big picture so as to better understand their teams, and how they can facilitate the closer integration of software development and live operations that is at the heart of the agile-focused DevOps approach.
I was reminded of something from 2 years back. Speaking at a Retail Technology conference Chris Howell (then IT Director eCommerce at Dixons Stores) talked about ‘Uniting the Tribes‘ within an organisation. Specifically he meant using monitoring of the user experience on the live site as the common language of data for the organisation that determines whether the tech guys deliverables are fit for purpose.
Of the presentations ‘DevOps For Dinosaurs‘ by Niek Bartholomeus was particularly good. Moving a large organisation to new, faster ways of working takes some people skills, and the ability to decide which new practises can wait until later, and which the organisation is ready for first.
Overall, the discussions were first rate. There were two or three hundred folks there! I guess the fact they’d booked a conference that straddled across to a Saturday meant that they were highly motivated, proactive guys!
The Monitor Everything discussion open space highlighted the fact that tech guys, no matter what flavour, do want more data to feed their analytical inquisitiveness! However, in reality, not all monitor data is equal: the low hanging fruit is the data that is closest to the hardware but sadly furthest from the end goal – the customers experience online.
The extra effort to monitor close to the customer was for some too much effort (hard line techies!) and for others was a goal they had in mind, but progress was hard. This reflected a lot of my experience doing website monitoring each day: the most useful, actionable monitoring is from ‘doing what the customer does’.
The Log Everything session also appealed to the need for data to troubleshoot. A good part of the discussion was questions being put to some of the guys who have been using logStash in earnest. It’s a tool we’ve just started looking at and it makes a convenenient, easily searchable, ‘bucket’ to put all log data into: from server hardware level, right up to application level unique logs.
For some, just the power as a troubleshooting tool, the ease of with which you can build powerful filters, was the main win. A couple of advanced users, though, were starting to log actual data so that they could track a user’s transaction through various systems with relative ease.
I must have a chat with our guys here and see how deep we have gone with logStash.