Just over a year ago we evolved the Central architecture to be globally load balanced with 2 servers in the US and 2 more in the UK. This year, we've gone even futher to increase reliability and delivery performance.
We evaluated several options and ultimately settled with Edgecast as the delivery system. The way the system works is we direct the central dns queries (repo1, uk, etc) to an Edgcast CNAME. This resolves to the fastest available server based on your location (this is effectvely what we had been doing previously with Dyn). The Edgecast network is serving as a worldwide proxy to serve up the content. If the content isn't available on the edge, it falls back to a middle tier and finally back our our root Central servers in the US or UK and then the content is cached along the way. In this way, the infrastructure backing Central and managing the inbound syncs, and indexes still functions as it did prior to this change. There's nothing you need to do to take advantage of the changes, it's already in place and hopefully the only side effect is faster builds. There's more info available on my blog: http://www.sonatype.com/people/2012/07/we-just-kicked-central-performance-and-availability-up-a-notch-with-edgecast/ Thanks, Brian