Facebook today announced that its newest data center in Altoona, Iowa, is now open for business. The new facility complements the company’s other centers in Prineville, Ore; Forest City, N.C. and Luleå, Sweden (the company also operates out of a number of smaller shared locations). This is the first of two data centers the company is building at this site in Altoona.
What’s actually more interesting than the fact that the new location is now online is that it’s also the first data center to use Facebook’s new high-performance networking architecture. Traditional data centers are, for the most part, built around clusters of machines. That’s because you need really high networking performance between all of the machines within a cluster.
Networking resources between clusters, however, are typically more limited. That’s fine, as long as your applications mostly run within a single cluster, but as Facebook grew, its applications also expanded beyond these single clusters. So the balancing act then is to correctly allocate resources for cluster-to-cluster communication and intra-cluster networking.
With Facebook’s new approach, however, the entire data center runs on a single high-performance network. There are no clusters, just server pods that are all connected to each other. Each pod has 48 server racks — that’s much smaller than Facebook’s old clusters — and all of those pods are then connected to the larger network. You can find all the sordid details about how exactly Facebook architected this setup here (and you can read all about the BGP4 routing protocol, TOR uplinks, as well as RIB and FIB resources there, too).
What matters, though, is that this is a very modular approach that has already allowed Facebook to increase its intra-building network capacity ten-fold compared to the old design and the company believes that a 50x improvement is possible.
And with all of that, your friends will get even faster alerts when you like their babies’ pictures on Facebook.