Re^2: Enhancing Speed

How much CPU time do you expect to use to gzip a 70KB page ?

I tried to gzip a 200KB file on my Core2Duo laptop ... it toke less than 1 second. Beside when you take 1 sec to deliver a 10KB page instead of 70KB which will load faster for the client and release the apache child to serve someone else I'm sure you'll be saving CPU time.

Comment on Re^2: Enhancing Speed

Replies are listed 'Best First'.
Re^3: Enhancing Speed by marto (Cardinal) on Aug 18, 2010 at 17:07 UTC
"How much CPU time do you expect to use to gzip a 70KB page ?" Wouldn't that depend on various factors? My point was that your suggested ten minute change could end up doing more harm than good, as others have also suggested. The metric in your example does not take into consideration any of the factors including the servers CPU speed, the servers memory, the applications (Perlmonks custom version of everything and the DB queries) or the load. As you say "I don't have the statistics of your site, nor the server specifications". Your suggestion that compressing data on your laptop (with a relatively modern CPU) is a comparable benchmark for running compression on old servers isn't, IMHO, a sensible approach to benchmarking the impact nor justification for someone taking ten minutes out of their day to implement your untested change, which may take longer to recover from should the resulting performance be unfavourable. Failure to consider your environment can lead to trouble.	[reply]
Re^4: Enhancing Speed by ahmad (Hermit) on Aug 18, 2010 at 17:56 UTC
Your suggestion that compressing data on your laptop (with a relatively modern CPU) is a comparable benchmark for running compression on old servers isn't I've just tested it on an old box I have a Celeron 1.3 GHZ CPU (an outdated box bought 8 years ago) with WinXP on it and a bunch of programs running and it toke me less than 1 second to compress 200 KB file. someone taking ten minutes out of their day to implement your untested change, which may take longer to recover from should the resulting performance be unfavourable. As I have stated, recovery is as easy as commenting off 1 line and restarting Apache ... and I've never seen anyone able to test changes without implementing it in the first place.	[reply]