Automatic Identification of Application I/O Signatures from Noisy Server-Side Traces on FAST 2014.

Competing workloads on a shared storage system cause I/O resource contention and application performance vagaries.

Running multiple enterprise workloads, such as Financial, MSN and Exchange, on a SSD concurrently will cause the Garbage Collection Interference. These enterprise workloads is hard to identified using black box methods.

On the other side, I/O-intensive application’s samples show certain repeated burst I/O patterns. For example, the IOR benchmark generate a quiet Spider storage system partition.

This paper focus on how to extract the target I/O signature when the noisy is too heavy that you get a sample like this:

The technology used is Wavelet transform domain filters and a grid-based clustering approach called CLIQUE.