Daniel,
> On 24 Apr 2020, at 12:35, Daniel Turczański <[email protected]> wrote:
>
> Thanks. Are there any benchmarks available for Pharo HTTP servers? Is it
> possible to achieve a performance similar to node.js? I know the biggest
> selling point of Smalltalk is not perf but still I'm curious.
Benchmarking is a black art with many pitfalls.
Let's say that Pharo HTTP performance is acceptable.
Here is a test that I just ran (on a LXC/LXD container on a big production
server).
t3@t3-pharo-test:~/tmp/pharo9$ cat /etc/lsb-release
DISTRIB_ID=Ubuntu
DISTRIB_RELEASE=18.04
DISTRIB_CODENAME=bionic
DISTRIB_DESCRIPTION="Ubuntu 18.04.4 LTS"
t3@t3-pharo-test:~/tmp/pharo9$ curl get.pharo.org/64/90+vm | bash
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 3054 100 3054 0 0 198k 0 --:--:-- --:--:-- --:--:-- 198k
Downloading the latest 90 Image:
http://files.pharo.org/get-files/90/pharo64.zip
Pharo.image
Downloading the latest pharoVM:
http://files.pharo.org/get-files/90/pharo64-linux-stable.zip
pharo-vm/pharo
Creating starter scripts pharo and pharo-ui
t3@t3-pharo-test:~/tmp/pharo9$ ./pharo Pharo.image printVersion
[version] 'Pharo9.0.0'
'Pharo-9.0.0+build.252.sha.fe2815c89f309789cbfba9a1ce821fb959dfe987 (64 Bit)'
t3@t3-pharo-test:~/tmp/pharo9$ ./pharo Pharo.image eval --no-quit 'ZnServer
startDefaultOn: 1701' &
[1] 32203
a ZnManagingMultiThreadedServer(running 1701)
t3@t3-pharo-test:~/tmp/pharo9$ curl http://localhost:1701/random
0BFE44ECF0BEEDDD949A8D5A363E5E6192A4B7CD4939E8BD7A7C7EC69995157
t3@t3-pharo-test:~/tmp/pharo9$ ab -k -n 1024 -c 8 http://localhost:1701/random
This is ApacheBench, Version 2.3 <$Revision: 1807734 $>
Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/
Licensed to The Apache Software Foundation, http://www.apache.org/
Benchmarking localhost (be patient)
Completed 102 requests
Completed 204 requests
Completed 306 requests
Completed 408 requests
Completed 510 requests
Completed 612 requests
Completed 714 requests
Completed 816 requests
Completed 918 requests
Completed 1020 requests
Finished 1024 requests
Server Software: Zinc
Server Hostname: localhost
Server Port: 1701
Document Path: /random
Document Length: 64 bytes
Concurrency Level: 8
Time taken for tests: 0.246 seconds
Complete requests: 1024
Failed requests: 0
Keep-Alive requests: 1024
Total transferred: 256000 bytes
HTML transferred: 65536 bytes
Requests per second: 4155.22 [#/sec] (mean)
Time per request: 1.925 [ms] (mean)
Time per request: 0.241 [ms] (mean, across all concurrent requests)
Transfer rate: 1014.46 [Kbytes/sec] received
This is a small, do nothing request, tested locally, and so on.
I would say it is safe to say you can handle 100s to 1000s of requests per
second on a single Pharo instance (which is using only a single core).
The good thing about HTTP is that you can normally scale horizontally by adding
more and more worker instances under a load balancer -- these are standard
techniques.
HTH,
Sven
> The Concurrency book [1] says that Pharo's Process is like a fibre and is
> lightweight. How big is the stack? Java Quasar's fibers are very lightweight
> with a 400 byte stacks [2]. Can you have hundreds or thousands of processes
> running blocking IO without much memory and performance penalty?
>
> [1] http://books.pharo.org/booklet-ConcurrentProgramming/
> [2]
> https://docs.paralleluniverse.co/quasar/javadoc/co/paralleluniverse/fibers/Fiber.html
>
>
>
> --
> Sent from: http://forum.world.st/Pharo-Smalltalk-Users-f1310670.html
>
