In this test there are other problems which maybe unfriendly to Nginx.
(1) Nginx will read the file from DISK for EVERY request. Again the
sendfile is disabled. Worse！ Although OS will cahed this file but KERNEL
-> Nginx Process Space will still cost a lot.
(2) Both Ring-Jetty and Http-kit just write the memory string to every
request because the file has been load into Clojure var "response".
See the code : "
(def response {:status 200 :headers {"content-type" "text/html"}
> :body (slurp "resources/index.html")
> })
>
> ;;;; Handlers
>
> (defn handler [request] response)
>
"
On Monday, January 13, 2014 12:10:40 PM UTC+8, Xfeep Zhang wrote:
>
>
> If reduce worker_connections from 20000 to 1024 or 128, I think nignx will
> be the fastest one in this test (the max concurrent level is only 96).
>
>
>
> On Saturday, January 5, 2013 11:52:57 PM UTC+8, Peter Taoussanis wrote:
>>
>> Hi all,
>>
>> Quick post to mention that I've put up some rough benchmarks for a number
>> of Clojure web servers:
>> https://github.com/ptaoussanis/clojure-web-server-benchmarks.
>>
>> Getting semi-reliable numbers was a real PITA, and I'm sure there's still
>> plenty of room left for improvement (suggestions welcome!). Usual
>> disclaimers apply: the numbers vary by hardware, by OS, by concurrency, and
>> by the response size (among other things). The servers also vary
>> considerably in their feature sets. Not to mention that your web server is
>> seldom going to be the bottleneck when generating dynamic content. Tl;dr:
>> this is all very difficult to generalize: you'll need to bench in your own
>> environment with your own workloads to get an accurate picture.
>>
>> Anyway I think the numbers are interesting (and pretty impressive
>> all-round). Nginx can almost certainly be tweaked faster (ideas?), but I
>> wouldn't have expected something like the standard Ring Jetty adapter even
>> to be playing in the same ballpark - so that's nice.
>>
>> I'll try keep this updated as servers are updated. And as I've mentioned
>> on the GitHub page, pull-requests welcome for anything I may have missed,
>> including other servers or mis-configurations, etc. Hopefully with some
>> tweaking we can converge on some reasonably accurate/useful common-case
>> numbers.
>>
>> Cheers!
>>
>
--
--
You received this message because you are subscribed to the Google
Groups "Clojure" group.
To post to this group, send email to clojure@googlegroups.com
Note that posts from new members are moderated - please be patient with your
first post.
To unsubscribe from this group, send email to
clojure+unsubscr...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/clojure?hl=en
---
You received this message because you are subscribed to the Google Groups
"Clojure" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to clojure+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
