The storage cluster initial nodes have been built up now and we have been doing extensive testing with it.
We've worked on the software side of things, and finally we have found a way to deliver both performance and flexibility, and bigger scale stability testing will ensue soon.
We've managed to get raw device performance over the network without any special "hacks" etc. and in a stable manner, infact, with the new software on the backend some nodes has been working and in use for a while now.
We are starting now to build up the arrays, which are initially going to be 12Tb each.
We had to downgrade our redundancy level significantly to get the performance required for this to be pulled off, but we will increase the same redundancy level by doing smaller pools, so if a pool is lost it will affect a maximum of 20 servers or so.
Eventually we will also provide optional backup services on a per node basis.
There will be a maintenance starting today as we upgrade one of the SAN gateways to faster network access. The time when this is done is not exactly known, but the maintenance is done so that we will not need to do any maintenance on that particular machine for a long time to come. This will affect a fraction of the nodes running currently.
Monday, August 5, 2013
