Page Comparison

...

1. Number of rpcs
2. Internal Hash-table size
3. Actual DRAM used versus raw DRAM required for object contents

Graphs of interest

...

Round 3 - (generated with code version - commit db388ad3)

• Co-ordinator and Server are both running on rc01
• Client making timing measurements is at rc03 - the queen
• Load - additional clients are run on rc04-6. If load > 3 is required, multiple clients are run on the same node. The Workers.
• Builds are with DEBUG=no unless otherwise noted.
• Object size is fixed at 1000 bytes.

...

• transport=fast+udp, Operations per second (throughput) versus load (number of running clients).
  

...

Round 2 - (generated with code version - commit 7a861e)

• Co-ordinator and Server are both running on rc01
• Client making timing measurements is at rc02
• Load - additional clients are run on rc03-8. If load > 6 is required, multiple clients are run on the same node.
• Builds are with DEBUG=no unless otherwise noted.
• Object size is fixed at 1000 bytes.
• Each "load" client is another instance of Bench, each running ten times as many operations as the measuring Bench client with continuous operations)

...

• DEBUG=yes, transport=fast+udp, Operations per second (throughput) versus load (number of running clients).
  

...

Round 1 - (generated with code version - commit b6af37)

• Time taken for reads (avg latency in ns) versus object size.
• Time taken for writes (avg latency in ns) versus object size.
  (avgns times are as measured at client side. avgctr times are as measured at server using the serverCounter.)
  

...