Pianola: A script-based I/O benchmark

Framework for Analyzing Android I/O Stack Behavior: From Generating the Workload to Analyzing the Trace

“…There are a number of works that address the limits of the synthetic workload [17][18][19][20][21]. These works exploit two approaches in capturing and replaying the system calls of an application.…”

“…These works exploit two approaches in capturing and replaying the system calls of an application. First approach is to capture the system calls at user-level and replaying them with timing accuracy [17,[19][20][21].…”

“…Mobigen records system calls using strace [26] and processes them to be replayed by Mobibench. There have been a number of works on capturing and replaying the I/O traces [17][18][19][20][21]. To the best of our knowledge, our work is the first to attempt capturing as well as replaying the I/O trace of Android applications.…”

Predicting disk I/O time of HPC applications on flash drives

“…Related work has pursued I/O modeling and prediction by using script based benchmarks to replay an applications causal I/O behavior [10,11] or using parameterized I/O benchmarks [12] to predict run time on the target system. Our modeling approach differs from the related work by using parameterized benchmarks to compute speedup ratios on target systems for each call and use that to predict an applications I/O time.…”

“…Pianola [10] is a script based I/O benchmark that captures causal information of I/O calls made by a sequential application. The information is captured by a binary instrumenter that, for each call, captures wall clock time of the call, the time spent servicing the call, and arguments passed to the call.…”