Since the cluster failure last month we had been running our rercursive name servers in a sub optimal (but still redundant) configuration. Last Friday, we completed the re-deployment and once again have a multiple layers of redundancy with both our primary and secondary recursive name servers announced into our network in Santa Rosa, San Francisco and Los Angeles. The old Alteon hardware load balancers that used to handle 208.201.224.11 in Santa Rosa and San Francisco have been replaced with Linux LVS/Quagga BGP servers that do not suffer from the same limitations that caused the original cluster failure. While each of the real servers could handle many times our total DNS load on its own, neither of the Alteons that were responsible for directing traffic for 208.201.224.11 could handle it after some routine network configuration shifted traffic inside of our network. Apart from a small hiccup due to a BGP misconfiguration Friday morning, the new clusters have performed perfectly. –Kelsey, Nathan and Jared