[Koha-devel] "The Plack effect" benchmarked

[Paul Poulain]

Hi,

Just in case some of you have doubts about the interest of having Plack
for staff interface, here is a result of
misc/load_testing/benchmark_staff.pl made with and without Plack:

benchmarking WITH Plack:
Step 1: staff client main page     	4178ms	4.786 pages/sec
Step 2: catalog detail page        	9991ms	2.001 biblios/sec
Step 3: catalogue search               	5631ms	3.551 biblios/sec
Step 5: patron detail page         	7037ms	2.842 borrowers/sec
Step 5: patron search page             	11800ms	3.389 borrowers/sec
Step 6a circulation (checkouts)        	8967ms	2.23 checkouts/sec
Step 6b circulation (checkins)         	13880ms	1.44 checkins/sec
all transactions at once               	55433ms	3.247 operations/sec
(note = tests run with plackup, not with starman, the all transaction at
once results would probably be better with starman)

benchmarking WITHOUT Plack:
Step 1: staff client main page     	28576ms	0.699 pages/sec
Step 2: catalog detail page        	39106ms	0.511 biblios/sec
Step 3: catalogue search               	55524ms	0.36 biblios/sec
Step 5: patron detail page         	34610ms	0.577 borrowers/sec
Step 5: patron search page             	64964ms	0.615 borrowers/sec
Step 6a circulation (checkouts)        	37907ms	0.527 checkouts/sec
Step 6b circulation (checkins)         	39129ms	0.511 checkins/sec
all transactions at once               138298ms 1.301 operations/sec

It appears that the speed improvement is more x3 or x4 than x1.5 as I
usually say
-- 
Paul POULAIN
http://www.biblibre.com
Expert en Logiciels Libres pour l'info-doc
Tel : (33) 4 91 81 35 08