[Date Prev][Date Next][Thread Prev][Thread Next][Minivend by date
][Minivend by thread
]
RE: mod_perl
> how are you benchmarking? what program are you using?
I am using the apache benchmark program under linux with this command:
#/usr/sbin/ab -n 150 -c 30 http://www.host.com/cgi-bin/store
so it does 150 requests total, and keeps 30 concurrent at all times.
Jay
>
> -----Original Message-----
> From: owner-minivend-users@minivend.com
> [mailto:owner-minivend-users@minivend.com]On Behalf Of Jay Austad
> Sent: Friday, June 18, 1999 9:19 AM
> To: minivend-users@minivend.com
> Subject: Re: mod_perl
>
>
> ****** message to minivend-users from Jay Austad <austad@fallon.com>
> ******
>
>
>
> > That's not right. Have you had MV build an INDEX for your products db?
>
> How do I do that?
>
> I am not swapping. When I do a free, it only shows that around 117MB of ram
> is being used, and no swap.
>
>
>
> > Actually, the great thing about MV is that it doesn't spawn a perl process
> > for each request. It runs one main perl process that waits for requests
> from
> > either vlink or tlink.
>
> I'm not so sure this is true. When I run apache benchmark against minivend,
> and watch 'top' on the server, it shows about 15 copies of perl running, and
> they are all taking up a good chunk of the cpu. In total, they take up
> about 180% of the CPU (dual processor). I assume the vlink program is
> simply just calling perl somehow.
>
> The machine has the latest versions of all software on it, all compile with
> pentium optimizations. Perl is what is running the load up so high on the
> server (it did the same thing before I compiled the latest version with
> optimizations). 30 concurrent requests shouldn't bring the load up much
> past 1 or so, but it goes above 10. 31 seconds on average to return a page
> under that load. Otherwise, with no load, it works fine, returns pages
> within 1 second or so.
>
> Ok, just ran a quick test. I ran the benchmark against /cgi-bin/store
> before, and that was the test that was running the load up to 10.72. I just
> ran apache benchmark again with 150 requests, 30 concurrent, against
> /cgi-bin/store/frames, and /cgi-bin/store/noframes, and then again on
> /cgi-bin/store.
>
> The results are below, first the frames, then noframes, and finally just
> /cgi-bin/store. The frames and noframes options are acceptable, but the
> last one is where it's really loading the server down. Isn't /cgi-bin/store
> the same thing as /cgi-bin/store/noframes? They return the same thing. So
> what extra stuff is happening when I don't give the frames or noframes
> option?
>
> Here are the results of the benchmark using the frames option:
>
>
> Server Software:
> Apache/1.3.6
> Server Hostname:
> www.host.com
> Server Port:
> 80
> Document Path:
> /cgi-bin/store/frames
> Document Length:
> 1371 bytes
> Concurrency Level:
> 30
> Time taken for tests:
> 45.844 seconds
> Complete requests:
> 150
> Failed requests:
> 0
> Total transferred:
> 235650 bytes
> HTML transferred:
> 205650 bytes
> Requests per second:
> 3.27
> Transfer rate:
> 5.14 kb/s received
> Connnection Times (ms)
>
> min
> avg
> max
> Connect:
> 16
> 66
> 3040
> Processing:
> 741
> 8205
> 9046
> Total:
> 757
> 8271
> 12086
>
>
>
> Here are the results using noframes:
>
> Server Software:
> Apache/1.3.6
> Server Hostname:
> www.host.com
> Server Port:
> 80
> Document Path:
> /cgi-bin/store/noframes
> Document Length:
> 2 bytes
> Concurrency Level:
> 30
> Time taken for tests:
> 43.238 seconds
> Complete requests:
> 150
> Failed requests:
> 0
> Non-2xx responses:
> 150
> Total transferred:
> 41550 bytes
> HTML transferred:
> 300 bytes
> Requests per second:
> 3.47
> Transfer rate:
> 0.96 kb/s received
> Connnection Times (ms)
>
> min
> avg
> max
> Connect:
> 18
> 66
> 3028
> Processing:
> 661
> 7776
> 6480
> Total:
> 679
> 7842
> 9508
>
> Here are the results of just calling /cgi-bin/store:
>
> Server Software:
> Apache/1.3.6
> Server Hostname:
> www.host.com
> Server Port:
> 80
> Document Path:
> /cgi-bin/store
> Document Length:
> 16947 bytes
> Concurrency Level:
> 30
> Time taken for tests:
> 172.749 seconds
> Complete requests:
> 150
> Failed requests:
> 0
> Total transferred:
> 2572050 bytes
> HTML transferred:
> 2542050 bytes
> Requests per second:
> 0.87
> Transfer rate:
> 14.89 kb/s received
> Connnection Times (ms)
>
> min
> avg
> max
> Connect:
> 17
> 115
> 3013
> Processing:
> 7139
> 31040
> 36508
> Total:
> 7156
> 31155
> 39521
>
>
> I have the PageCache option set. Is it caching the frames and noframes
> options and not caching the page when I call just the store cgi? I'd really
> like to figure out why this is happening and fix it instead of just making
> everyone go to /cgi-bin/store/noframes instead.
>
>
>
> Jay
> -
> To unsubscribe from the list, DO NOT REPLY to this message. Instead, send
> email with 'UNSUBSCRIBE minivend-users' in the body to
> Majordomo@minivend.com.
> Archive of past messages: http://www.minivend.com/minivend/minivend-list
>
> -
> To unsubscribe from the list, DO NOT REPLY to this message. Instead, send
> email with 'UNSUBSCRIBE minivend-users' in the body to Majordomo@minivend.com.
> Archive of past messages: http://www.minivend.com/minivend/minivend-list
