Dan’s View

Well not espically flavoursome alerts, nothing critical. The day kinda just passed without too many fires. I think we even managed to keep the number of active alerts under 30! Which is a huge improvment upon a few weeks ago. I can see that we will eventually reach a point we the number is so low that we can trust each alert and start to SMS alert, and therefore not have to watch a web
interface to the alert system.

In talking with one of my collueages a while back about what they used at his company they actually reached the point where there is no manual deployment of servers. The system knows the load/peaks/demand for the services and if a node falls over it automagially re-images it, well if the hardware is borked it would re-purpouse another machine to take it’s role. I have been thinking
about this for a while and it would be an extreamly efficent way of handling the whole kaboodle that is the chain that has to occur in the case of new hardware being provisioned or the re-purposing of existing. It can cause headaches, espcially if someone accidently makes a typo in the Bind
configuration; that has been know to take the site offline before… It is something that I would like to possibly start working towards; although is a massive project, at least if feels massive. If I start to break the task down into chunks I am sure I could manage it. Let’s see.

Too tired to join colluages on a meal tonight; it’s a shame as some of the folk from the US offices are over at the moment. So the evening looks like Mario Kart; I am turning into a bad bad Sys. Admin who plays games in the evening. Still they are fun.

Acton Town stop, I think I have written enough now. Have a good evening. Oh and Happy Birthday Jens!

Cheers,
DanB