Jump to content

Obsolete:Data Dump Redesign/Throughput

From Wikitech

This page is kept for historical purposes only.

Stubs

  • 200K Samples
    • ESWIKI
      • real 20m19.417s / 25M
      • user 13m40.800s
      • sys 0m31.240s
      • ESWIKI has just under 1.2million entries
      • 1,170,721 as of 2009 06 15 Pages
      • Currently takes a little more then an hour to dump stubs
    • ENWIKI
      • 17,194,115 Pages
      • Split a the stubs from ENWIKI 20090604 into 200k chunks and process
      • 100Stubs @ 8core = 13 Machines

Splitting Stubs

Wed Jun 24 00:50:25 2009 Processed 10000 pages out of 200000 for segment 0
Wed Jun 24 01:18:41 2009 Closing Segment 0 after 200000 total pages processed. Opening next segment
  • 30Mins per stub