Changes between Version 9 and Version 10 of YumBenchmarks

Show
Ignore:
Author:
james (IP: 65.172.155.230)
Timestamp:
09/29/10 16:05:30 (7 years ago)
Comment:

Add install/remove benchmarks, after recent work.

Legend:

Unmodified
Added
Removed
Modified
  • YumBenchmarks

    v9 v10  
    260260||repoquery yum+yumdb||0.421|| 
    261261||yum list installed yum||0.642|| 
     262 
     263== Comparison between latest "yum install/remove", rpm == 
     264 
     265For a long time yum had been "much worse" than rpm for install/remove commands. Some of that was due to general fixes, but it also generally has got less love because any difference in speed was only really noticeable with small numbers of small packages (otherwise the I/O from the install/remove vastly outweighs any other part of the operation). 
     266 
     267It is also worth noting that recent versions of yum also have to do a non-trivial amount more work than plain rpm, as yum keeps a detailed history log and it's own yumdb, so yum should never as fast as rpm ... but we can (and do) endevour to get close. 
     268 
     269These timings were performed on a virt-machine: 
     270 
     271||command||version||operation||time (seconds)|| 
     272 
     273 
     274||rpm||4.8.1-2||install||2.0|| 
     275||rpm||4.8.1-2||remove||1.5|| 
     276 
     277||yum||3.2.25-1||install||7.0|| 
     278||yum||3.2.25-1||remove||5.5|| 
     279 
     280||yum||3.2.28-3||install||6.5|| 
     281||yum||3.2.28-3||remove||4.0|| 
     282||yum||3.2.28-8||install||4.0|| 
     283||yum||3.2.28-8||remove||3.0|| 
     284||yum||3.2.28-8||localinstall||3.2|| 
     285||echo no | yum||3.2.28-3||install||2.5|| 
     286||echo no | yum||3.2.28-3||remove||1.5|| 
     287||echo no | yum||3.2.28-8||install||1.5|| 
     288||echo no | yum||3.2.28-8||remove||0.5|| 
     289||echo no | yum||3.2.28-8||localinstall||0.6|| 
     290 
     291...note that for all the operations except the "echo no" variants, there could be a non-trivial delay due to IO stalls.