In our operations engineering group here at AdMob these are the numbers that we live by.
We aspire to provide our customers a great experience with our products and services every hour of every day without interruption. In order to fulfill these aspirations we focus a significant amount of our total engineering capacity on building fast, speedy infrastructure that can handle ever increasing traffic volumes.
From 2007 to 2008 the number of ad requests we received increased by a factor of 3.7 year-over-year to 52.4 billion. To keep up with this growth we’ve deployed dozens of significant infrastructure changes over the past 12 months. These investments in infrastructure have also allowed us to improve the quality of our service by increasing the speed with which we respond to ad requests, reducing the number of seconds per month that our services are unavailable, and increasing the speed of admob.com.
Given the complexity of some of these infrastructure changes (think of changing a tire on a car moving at 60 mph) and in spite of careful planning on our part, sometimes we do have hiccups. We experienced two such hiccups this week, one on Tuesday afternoon as we were making a series of changes to our global networking infrastructure and one on Wednesday afternoon as we were applying a set of changes to our primary databases. On Tuesday our services were partially available for approximately 45 minutes and on Wednesday users were unable to log in to admob.com for a similar length of time. We apologize to all of our customers for any inconvenience.
The good news is that following these two unfortunate service interruptions our infrastructure is now better than ever resulting in a faster AdMob for all of our customers. Even though we never plan for service interruptions of this nature and don’t anticipate any in the future, we invite all of you to follow us on Twitter (http://twitter.com/admob) where we’ll update customers about any future in-progress operational issues with admob.com.
Kevin Scott
VP Engineering
