Cloud Computing Update
Recently we made a post about our new cloud computing infrastructure and the hardware we were using to build such a platform. Well, our servers have been delivered and we are in stage two of this set-up which is the install and configuration of the ‘cloud’.
As previously discussed, we have purchased a number of the new Dell PowerEdge R710 servers, and we have since configured these in the data centre and commenced the complex process of getting these servers working in unison. The reason why this stage is so important, is because this is the base building blocks of our cloud. These machines will be doing the hard work of computing the extreme amounts of data we intend on pushing through the cloud. They will also be configured so that in the case that one server (Server A) was to fail or go offline, there would be no downtime for the end user, The other servers (Server B, C & D) in the cloud would compensate and manage the tasks that server A was doing. The data would autonomously be mirrored across the active servers, in the case that another server was to fail. The way that the servers do this, is via a backbone gigabit network.
The servers are connected to two different networks. Network A is the local network, purely for the servers to communicate to each other, exchange data and make sure they are all on the same page. Whilst network B is a public network, purely for internet access, So that the servers can speak to the world and give customer connectivity to their applications or virtual servers within the cloud.
During this time we will be purposely discconecting servers and in a sense, trying to do as much “normal” damage as possible to the cloud computing platform. What we meen by this is, We will be testing the redundancy of the cloud by simulating things like server crashes, network outages and software failure. This will allow us to test and if needed, do further tweaking on the configuration of the server or network to ensure our new cloud will truely be able to provide our customers with maximum uptime and redundancy. (We’re aiming for a 100% uptime guarunteed SLA).
Here are a couple of photos of the initial testing environment we have set up.
Related posts:
