Our latest arrival in July 2016 is this 1U server by HP that is destined to be our third VM host in our cluster environment. Running a cluster allows us to build fault-tolerant and robust systems, e.g. we can implement quorums in services and can thus prevent dangerous situations like split brains.
Regarding the specs this server is a bit less powerful compared to its neighbors: two Intel Xeon L5520 Quad-Core CPUs are served by 32GB RAM. In total we have given the server four 146GB SAS hard drives that are joined to a RAID5.
Just like any other server we wanted to focus on both redundancy and performance. To prevent the network load from being a bottleneck, we added two additional GBit NICs to the two interfaces on-board. Two of these interfaces each will form a trunk (LACP) and are routed as a cable bunch to a switch.
The other pair will be connected to another switch in the future. Using both local trunks, one can implement a failover interface and enable a redundant network connection.