Quantifying Temporal and Spatial Quantifying Temporal and Spatial - PowerPoint PPT Presentation

Quantifying Temporal and Spatial Quantifying Temporal and Spatial Localities Localities Florida State University fox, lojpur, awang@cs.fsu.edu Florida State University, {fox, lojpur, Florida State University, {fox, lojpur, 1 1 awang@cs.fsu.edu} awang@cs.fsu.edu}

Quantifying Localitites Quantifying Localitites  Describe workloads Describe workloads  _ Compare different workloads Compare different workloads _ _ Create accurate synthetic workloads Create accurate synthetic workloads _  Understand locality transformations Understand locality transformations  _ How are workloads transformed by caching? How are workloads transformed by caching? _ _ Do workloads properly stress data path components? Do workloads properly stress data path components? _ _ What opportunities remain for further data path optimizations? What opportunities remain for further data path optimizations? _ Florida State University, {fox, lojpur, 2 awang@cs.fsu.edu}

Current Metrics Current Metrics _ Cache hit ratios Cache hit ratios _  Measures the effectiveness of caching Measures the effectiveness of caching   Does not cleanly separate temporal and spatial locality Does not cleanly separate temporal and spatial locality   Does not show transformations Does not show transformations   Only applies to cache Only applies to cache  _ Reference distance Reference distance _  Reference distance is the number of unique blocks referenced Reference distance is the number of unique blocks referenced  before accessing the same block before accessing the same block _ Block distance Block distance _  Block Distance is the difference between block locations on disk Block Distance is the difference between block locations on disk  Florida State University, {fox, lojpur, 3 awang@cs.fsu.edu}

New Metrics New Metrics _ Affinity Affinity _  Builds on block and reference distance Builds on block and reference distance   0%-100% easy to understand 0%-100% easy to understand   Less sensitive to generations of hardware Less sensitive to generations of hardware  Florida State University, {fox, lojpur, 4 awang@cs.fsu.edu}

Example use Example use Web-trace Synthetic Shuffled references, same inter-arrival rate, reference distribution Florida State University, {fox, lojpur, 5 awang@cs.fsu.edu}

Future Work Future Work _ Study how each storage component Study how each storage component _ transforms locality characteristics transforms locality characteristics _ Verify fidelity of various benchmarks Verify fidelity of various benchmarks _ _ Improve storage components and Improve storage components and _ benchmarks benchmarks Florida State University, {fox, lojpur, 6 awang@cs.fsu.edu}