You are here
Linux Systems Capacity Planning: Beyond RRD and top
As infrastructure costs rise, there's an urgent need to squeeze more performance from the same hardware. After several years of measuring and managing the capacity of thousands of Linux servers, we have learned that most typical tools and metrics are not sufficient to predict performance bottlenecks, particularly during traffic spikes. By using queue theory formulas and instrumenting our applications we were able to find the limits of our systems, improve reliability, and maximize throughput and performance.
Open Access Media
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.