Check out the new USENIX Web site. next up previous
Next: Storage Response Time Up: Web Service Models Previous: Server Memory

   
Storage I/O Rate

Given an estimate of cache hit ratio H, the service's storage I/O load is easily derived: for a Web load of $\lambda $ requests per second, the I/O throughput demand $\lambda _S$for storage (in IOPS, or I/O operations per second) is:


\begin{displaymath}\lambda_S = \lambda S (1 - H)
\end{displaymath} (2)

The average object size S is given as the number of I/O operations to access an object on a cache miss.



Ronald Doyle
2003-01-20