At Limelight, we pride ourselves on delivering excellent service, including a superior cache hit rate. High cache hit rates improve response time, throughput, and availability, and protect customer origins from high requests per second and high traffic. High cache hit rates help customers to avoid costly buildouts of networks, servers and locations to protect them, because Limelight provides that protection for them.
Through Limelight’s diligent and relentless efforts, we have made a phenomenal leap in cache at the edge. In fact, utilized cache at the edge is now 33 petabytes, which is 50% more than the 22 petabytes from just a few months ago.
What is the impact? One way to measure it is to look at the data from one of our customers, a multi-national software and device manufacturer. For this customer, we charted out cache hit ratio and requests per second from March 1-June 10, 2016.
Here is what we see:
In a time of significant CDN traffic growth for us and the industry, this customer had a cache hit rate of 98.45% on May 1, and by June 1 had moved to an incredible cache hit rate of 99.83%. Even better, the cache hit rate had very low variability (standard deviation of a mere 0.65%) even though their traffic was highly variable!
On any given day, we are managing an astonishing 20 to 25 billion objects in cache at the edge. We have added 50% more total cache at edge while significantly reducing the number of total servers as we move to be much more dense and green and power efficient. Proof that server count is an irrelevant metric, and is only a sign of inefficiency and an aging fleet!
We have the best CDN service, and now there is 50% more of it—with better performance, fewer total servers, less total electricity, and less total floor space. We’ve gone from great to incredible!