Feed Fiber

FeedBurner has reached a point where we will very soon begin sending out over 100Mb/second of feed content. Those of you playing the Ethernet home game (it's WANtastic!) will recognize 100Mb/sec as the upper limit on Ethernet traffic, so we have spent the last couple weeks preparing to upgrade to gigabit connectivity, which mostly involves paying more money for switches that look like the other switches that didn't cost so much money. Saturday, we spent a couple restful hours (and by "we", I mean "Joe and Paul") at our central hosting facility in Chicago swapping out Ethernet gear and replacing it with gigabit gear throughout the network. We were able to take our time swapping the gear in production, because of Feed Insurance, which we first described on January 5. While connectivity was interrupted in Chicago for only 1 minute total, we leveraged Feed Insurance, a second FeedBurner facility that served all feed traffic during that time. In order to both test Feed Insurance and upgrade to gigabit connectivity, we moved all the feed traffic to the second facility well in advance of the Chicago upgrade, watched the traffic migrate, connected the new gigabit gear in Chicago, and then brought the traffic back to Chicago. This was all transparent to feed subscribers, as our secondary facility was serving up the most recent copy of your feed. In its current configuration, Feed Insurance does not allow publishers to make feed configuration changes while running through the secondary facility. That's a logical next step for more thorough geo-redundancy. Publishers will see Feed Insurance bots polling their feeds. We will continue to expand our scalability and availability systems, and we may continue to test more Feed Insurance upgrades from time to time (additional remote facilities, geo-distributed feed serving, etc.), so next time you see the Feed Insurance message on our site, you'll know that Joe and Paul are spending another beautiful Saturday morning moving traffic around.