Our goal is continuous availability!
If you read my first blog, you will know that I talk a bit about multi sourcing. That our infrastructure is broken down into smaller pieces, each of them serviced by its own provider and Service Level Agreement.
The IT infrastructure can basically be broken down into; Hardware, Housing, Network, Daily Operation, Monitoring, Incident and Emergency Management. And then you have the Cloud services too, which provide another dimension to all these elements and hereby introduce the option to easy and fast scale our capability.
In order to provide continuous availability we will restructure and divide our own infrastructure and implement the best of each in the above mentioned areas. We have already identified some of the partners we would like build this structure with. It’s my wish that we are live on this new setup by the end of the year.
When we introduce many different service partners it naturally demands that we have the capability to interact and constantly evaluate them all in order to be true to the “best of breath” religion. So we are hiring! Check out our job section and you’re welcome to apply or send it to someone you know. Thanks J
We have built-in our own monitoring and performance tools into our application. Building your own tools may seem like an excellent idea in the beginning. However, you will be stuck with maintaining these tools that rarely are convenient in the priority together with the other important operational related tasks. As a result, they don’t follow with the development of your operation. Realizing this, we have now replaced our own monitoring system with a far more professional one, which gives us way more insides on to how our application servers are handling the different versions of applications code.
We are implementing Configuration Management. This gives us the ability to automatically build up new application servers – both on premise and in the cloud – and hereby quickly scale our capabilities. Furthermore, we will be able to deploy new features and bug fixes a lot faster.
In January this year, we replaced our database servers with two new big ones. However, they are still running on an older software version. This year we will upgrade that, giving us around 20% more performance. We have not done this before because we wanted to make sure it did not break anything. The recent tests indicate that we are good to go.
Possible service windows
We want to always be on, goddamn fast and available, no matter what you’re doing and how many are doing it at the same time.
Service windows do not go hand in hand with continuous availability. It’s more or less the opposite actually. However, to achieve our goals we have scheduled five service windows going forward for 2013 and 2014;
- June 8 2013 from 21:00 until June 9 2013 06:00
- Sep 7 2013 from 21:00 until Sep 8 2013 06:00
- Dec 7 2013 from 21:00 until Dec 8 2013 06:00
- March 29 2014 from 21:00 until March 30 2014 06:00
- June 7 2014 from 21:00 until June 8 2014 06:00
I promise that we will work like crazy and twist our brains to avoid using these service windows. If we need them, we will post the information here on the blog.