[Seaside] Pharo 1.1.1 and Cog?

[Lukas Renggli]

I measured roughly 3x speedup with latest Seaside build from hudson
and the original Kom Server. The test URL is the multi-counter
application with the first counter incremented to 2. The identical
image is used for both benchmarks.

  ab -n 1000 "http://127.0.0.1:8080/examples/multicounter?_s=WVGQTRo_5ZeqD-lv&_k=-P9AttB0aDi4r3B-"

Using concurrent requests (e.g. with "-c 10") decreases the speedup
(e.g. Cog is 2x faster).

Replacing the original Kom with a Kom that doesn't use DynamicBindings
increases the speedup slightly.

== with Cog ===========================================================

Server Software:        KomHttpServer/7.1.3
Server Hostname:        127.0.0.1
Server Port:            8080

Document Path:
/examples/multicounter?_s=WVGQTRo_5ZeqD-lv&_k=-P9AttB0aDi4r3B-
Document Length:        2435 bytes

Concurrency Level:      1
Time taken for tests:   2.178 seconds
Complete requests:      1000
Failed requests:        4
   (Connect: 0, Receive: 0, Length: 4, Exceptions: 0)
Write errors:           0
Total transferred:      2610004 bytes
HTML transferred:       2435004 bytes
Requests per second:    459.21 [#/sec] (mean)
Time per request:       2.178 [ms] (mean)
Time per request:       2.178 [ms] (mean, across all concurrent requests)
Transfer rate:          1170.44 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    0   0.1      0       1
Processing:     2    2   0.8      2      19
Waiting:        0    2   0.9      2      17
Total:          2    2   0.8      2      19

== without Cog ==========================================================

Server Software:        KomHttpServer/7.1.3
Server Hostname:        127.0.0.1
Server Port:            8080

Document Path:
/examples/multicounter?_s=iioGgbY2Pks-Qo9Q&_k=1D8QQU3FOzmM3-ia
Document Length:        2435 bytes

Concurrency Level:      1
Time taken for tests:   7.213 seconds
Complete requests:      1000
Failed requests:        4
   (Connect: 0, Receive: 0, Length: 4, Exceptions: 0)
Write errors:           0
Total transferred:      2610008 bytes
HTML transferred:       2435008 bytes
Requests per second:    138.63 [#/sec] (mean)
Time per request:       7.213 [ms] (mean)
Time per request:       7.213 [ms] (mean, across all concurrent requests)
Transfer rate:          353.34 [Kbytes/sec] received

Connection Times (ms)
              min  mean[+/-sd] median   max
Connect:        0    0   0.1      0       1
Processing:     6    7   3.3      7     106
Waiting:        0    7   3.3      7     101
Total:          6    7   3.3      7     107

=============================================================

On 12 October 2010 15:58, Johan Brichau <johan at inceptive.be> wrote:
> We just switched to Cog last week.
>
> All requests in our application (NeXTPLAN) run 5 times faster using Pharo1.1.1 and Cog.
>
> Needless to say: it's great work!!
>
>
> On 12 Oct 2010, at 09:32, Davorin Rusevljan wrote:
>
>> On Tue, Oct 12, 2010 at 9:27 AM, Lukas Renggli <renggli at gmail.com> wrote:
>> Sure, the one-click images run on Cog.
>>
>>
>> whoha, that is great!  How does it stack up regarding seaside performance when compared to classic VM? Any stability issues?
>>
>> rush
>> http://www.cloud208.com/
>>
>> _______________________________________________
>> seaside mailing list
>> seaside at lists.squeakfoundation.org
>> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside
>
> _______________________________________________
> seaside mailing list
> seaside at lists.squeakfoundation.org
> http://lists.squeakfoundation.org/cgi-bin/mailman/listinfo/seaside
>



-- 
Lukas Renggli
www.lukas-renggli.ch