Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
On Tuesday, January 21, 2014 6:25:02 AM UTC+8, Sergey Didenko wrote: Hi Xfeep, What are the good ways to handle some heavy Clojure calculations when using nginx-clojure? Do you mean every request will cost too much time ? If time cost is mainly caused by IO blocking, java thread pool can be used to resolve this problem. Otherwise you must add more computers or use more fast hardware to handle those pure CPU concentrated tasks. Under nginx model it's bad to block other incoming requests by holding a working thread for too long, right? Yes. Typically the nginx worker processes won't be too many and maybe the same number of your cpu cores. If you can not reduce single reponse time, all worker will be blocked by those slow tasks. So is it better to route complex job to http-kit? Or to use some kind of queue? Or may be to use this nginx-clojure JVM pool that is off by default? Suppose your time cost is mainly caused by IO blocking, java thread pool can be used by two ways : (1) use *pcall* , *pvalue* etc. provided by clojure to execute your tasks parallel to reduce single request-response time. OR (2) with nginx-clojure just simply set jvm_workers to some medium number eg 100, or bigger number if you get more memory. This thread pool will make nginx workers not blocked until all threads are exhausted. When jvm_workers 0, there's additional cost to one transfer event flag by pipe. But this cost can be ignored for your slow response and this cost will be lower than using nginx as a proxy to pass requests to backend java server. In the future release of Nginx-Clojure, synchronized non-blocking IO APIs will be provided. I think those synchronized non-blocking APIs will simpilfy to handle some IO blocked tasks. By that time jvm_workers maybe no need to be 0 . On Wed, Jan 15, 2014 at 3:18 PM, Xfeep Zhang easy...@163.comjavascript: wrote: On Wednesday, January 15, 2014 12:39:57 PM UTC+8, Roberto De Ioris wrote: i strongly suggest you to avoid the performance as a selling point, your project is cool but not for performance (and you are using a pipe to transfer requests data from nginx to the jvm so there ipc in place too, even if you can improve things using OS-specific syscall like splice). Although I have make it clear in the nginx english mail list, people joined clojure group may still misunderstand. So please forgive me repeating the message here. In the nginx english mail list , I have said : With the default setting pipe is not used. Pipe is only used for enable jvm thread pool mode only when jvm_workers 0 (jvm_workers default = 0). Further more pipe is never used to transfer the whole request or response message. When under jvm thread pool mode, pipe is only used to transfer a event flag (only one pointer size)。 -- Roberto De Ioris http://unbit.it -- -- You received this message because you are subscribed to the Google Groups Clojure group. To post to this group, send email to clo...@googlegroups.comjavascript: Note that posts from new members are moderated - please be patient with your first post. To unsubscribe from this group, send email to clojure+u...@googlegroups.com javascript: 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+u...@googlegroups.com javascript:. For more options, visit https://groups.google.com/groups/opt_out. -- -- 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.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
I see, thanks. On Tue, Jan 21, 2014 at 4:17 PM, Xfeep Zhang easyj...@163.com wrote: On Tuesday, January 21, 2014 6:25:02 AM UTC+8, Sergey Didenko wrote: Hi Xfeep, What are the good ways to handle some heavy Clojure calculations when using nginx-clojure? Do you mean every request will cost too much time ? If time cost is mainly caused by IO blocking, java thread pool can be used to resolve this problem. Otherwise you must add more computers or use more fast hardware to handle those pure CPU concentrated tasks. Under nginx model it's bad to block other incoming requests by holding a working thread for too long, right? Yes. Typically the nginx worker processes won't be too many and maybe the same number of your cpu cores. If you can not reduce single reponse time, all worker will be blocked by those slow tasks. So is it better to route complex job to http-kit? Or to use some kind of queue? Or may be to use this nginx-clojure JVM pool that is off by default? Suppose your time cost is mainly caused by IO blocking, java thread pool can be used by two ways : (1) use *pcall* , *pvalue* etc. provided by clojure to execute your tasks parallel to reduce single request-response time. OR (2) with nginx-clojure just simply set jvm_workers to some medium number eg 100, or bigger number if you get more memory. This thread pool will make nginx workers not blocked until all threads are exhausted. When jvm_workers 0, there's additional cost to one transfer event flag by pipe. But this cost can be ignored for your slow response and this cost will be lower than using nginx as a proxy to pass requests to backend java server. In the future release of Nginx-Clojure, synchronized non-blocking IO APIs will be provided. I think those synchronized non-blocking APIs will simpilfy to handle some IO blocked tasks. By that time jvm_workers maybe no need to be 0 . On Wed, Jan 15, 2014 at 3:18 PM, Xfeep Zhang easy...@163.com wrote: On Wednesday, January 15, 2014 12:39:57 PM UTC+8, Roberto De Ioris wrote: i strongly suggest you to avoid the performance as a selling point, your project is cool but not for performance (and you are using a pipe to transfer requests data from nginx to the jvm so there ipc in place too, even if you can improve things using OS-specific syscall like splice). Although I have make it clear in the nginx english mail list, people joined clojure group may still misunderstand. So please forgive me repeating the message here. In the nginx english mail list , I have said : With the default setting pipe is not used. Pipe is only used for enable jvm thread pool mode only when jvm_workers 0 (jvm_workers default = 0). Further more pipe is never used to transfer the whole request or response message. When under jvm thread pool mode, pipe is only used to transfer a event flag (only one pointer size)。 -- Roberto De Ioris http://unbit.it -- -- You received this message because you are subscribed to the Google Groups Clojure group. To post to this group, send email to clo...@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+u...@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+u...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out. -- -- 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. -- -- 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
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Hi Xfeep, What are the good ways to handle some heavy Clojure calculations when using nginx-clojure? Under nginx model it's bad to block other incoming requests by holding a working thread for too long, right? So is it better to route complex job to http-kit? Or to use some kind of queue? Or may be to use this nginx-clojure JVM pool that is off by default? On Wed, Jan 15, 2014 at 3:18 PM, Xfeep Zhang easyj...@163.com wrote: On Wednesday, January 15, 2014 12:39:57 PM UTC+8, Roberto De Ioris wrote: i strongly suggest you to avoid the performance as a selling point, your project is cool but not for performance (and you are using a pipe to transfer requests data from nginx to the jvm so there ipc in place too, even if you can improve things using OS-specific syscall like splice). Although I have make it clear in the nginx english mail list, people joined clojure group may still misunderstand. So please forgive me repeating the message here. In the nginx english mail list , I have said : With the default setting pipe is not used. Pipe is only used for enable jvm thread pool mode only when jvm_workers 0 (jvm_workers default = 0). Further more pipe is never used to transfer the whole request or response message. When under jvm thread pool mode, pipe is only used to transfer a event flag (only one pointer size)。 -- Roberto De Ioris http://unbit.it -- -- 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. -- -- 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.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
On Wednesday, January 15, 2014 12:39:57 PM UTC+8, Roberto De Ioris wrote: Hi Mingli, Thanks for your suggestion. Nginx-Clojure is quite different from uwsgi when supports JVM. Nginx-Clojure make JVM embed into Nginx worker process. JVM and Nginx worker have the same memory process space. Nginx-Clojure heavy uses pointer operation just like C to handle memory with Nginx worker. If I'm not wrong, uwsgi create every process for every request, or shared JVM processes between request? uWSGI creates a pool of processes with a number of threads in each and then they are used for the whole server life cycle. (a pretty standard behaviour) When using Nginx-Clojure there's no IPC cost even no thread switch cost if jvm_workers = 0 which is default. So it's why Nginx-Clojure is so fast! i strongly suggest you to avoid the performance as a selling point, your project is cool but not for performance (and you are using a pipe to transfer requests data from nginx to the jvm so there ipc in place too, even if you can improve things using OS-specific syscall like splice). It is cool because it simplify deployments, but nginx is not an application server, so things like multi-strategy gracefully reloads, stuck-requests managers, offloading and so on are not available. Nginx is not an application server, but it can be a good application with some module such as nginx-lua which largely used in production. Regarding the ipc problem, you generally put a load balancer on front of your app, so there is always ipc If we use F5 or LVS(Linux Virtual Server) as a balancer there 's no IPC :) Do not get me wrong, as i have already said your project is cool, but you should focus not on performance but on better integration with the nginx api for writing non-blocking apps (if possible), something at uWSGI unfortunately failed (the JVM has no support for switching stacks out of the box, something that is required for async mode in uWSGI) Thank you very much! Really good advices ! :) -- -- 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.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
On Wednesday, January 15, 2014 12:39:57 PM UTC+8, Roberto De Ioris wrote: i strongly suggest you to avoid the performance as a selling point, your project is cool but not for performance (and you are using a pipe to transfer requests data from nginx to the jvm so there ipc in place too, even if you can improve things using OS-specific syscall like splice). Although I have make it clear in the nginx english mail list, people joined clojure group may still misunderstand. So please forgive me repeating the message here. In the nginx english mail list , I have said : With the default setting pipe is not used. Pipe is only used for enable jvm thread pool mode only when jvm_workers 0 (jvm_workers default = 0). Further more pipe is never used to transfer the whole request or response message. When under jvm thread pool mode, pipe is only used to transfer a event flag (only one pointer size)。 -- Roberto De Ioris http://unbit.it -- -- 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.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
I have done the first one. The result is HEREhttps://github.com/ptaoussanis/clojure-web-server-benchmarks( https://github.com/ptaoussanis/clojure-web-server-benchmarks ) Thanks Taoussanis for his invitation to the project clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarkshosted on Github. On Tuesday, January 14, 2014 10:31:03 AM UTC+8, Xfeep Zhang wrote: You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzlehttps://launchpad.net/drizzlea no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.com wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request) (2) http-kit 2.1.16 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.691 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Hi, Thanks for your work on nginx-clojure. It looks great! As I know Nginx spawns many processes(correct me if I am wrong), does that mean, there will be many JVM process? On Tuesday, January 14, 2014 4:44:18 PM UTC+8, Xfeep Zhang wrote: I have done the first one. The result is HEREhttps://github.com/ptaoussanis/clojure-web-server-benchmarks( https://github.com/ptaoussanis/clojure-web-server-benchmarks ) Thanks Taoussanis for his invitation to the project clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarkshosted on Github. On Tuesday, January 14, 2014 10:31:03 AM UTC+8, Xfeep Zhang wrote: You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzlehttps://launchpad.net/drizzlea no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.com wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request)
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
You are welcome! Yes, you are right. One JVM instance is embed per Nginx Worker process. The number of Nginx Workers is generally the same with the number of CPU. If one Worker crashs the Nginx Master will create a new one so you don't worry about JVM crashs accidentally. Although there will be several JVM instances, there 's only one main thread attached with the Nginx Woker process. So the JVM instance uses less memory and no thread context switch cost in every JVM instance. In some cases If you can use only one JVM instance, you can set the Nginx Worker number to be 1 and set jvm_workers 1, nginx-clojure will create a thread pool with fixed number of thread. to handle requests for you. On Tuesday, January 14, 2014 5:50:34 PM UTC+8, Feng Shen wrote: Hi, Thanks for your work on nginx-clojure. It looks great! As I know Nginx spawns many processes(correct me if I am wrong), does that mean, there will be many JVM process? On Tuesday, January 14, 2014 4:44:18 PM UTC+8, Xfeep Zhang wrote: I have done the first one. The result is HEREhttps://github.com/ptaoussanis/clojure-web-server-benchmarks( https://github.com/ptaoussanis/clojure-web-server-benchmarks ) Thanks Taoussanis for his invitation to the project clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarkshosted on Github. On Tuesday, January 14, 2014 10:31:03 AM UTC+8, Xfeep Zhang wrote: You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzlehttps://launchpad.net/drizzlea no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.com wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Hi, Xfeep, Thanks for your contribution, and the project looks interesting. For me, the idea of driving ring webapp behind nginx is not new. We use uwsgi to drive our ring app behind nginx in our production. uwsgi support JVM and ring for almost one year, and I think the code is relative stable right now. - it support a native protocol between nginx and uwsgi which is more efficient than http - it support unix socket - and a rich uwsgi api layer to provide some means to communicate between webapps - and according to the performance tests by the author, it is a little bit faster than jetty. It is on our production for half a year, quite stable, and very harmonious with the python app. I am not want to sale the solution of uwsgi, but it worth taking a look and make some comparison. Regards, Mingli On Tue, Jan 14, 2014 at 9:12 PM, Xfeep Zhang easyj...@163.com wrote: You are welcome! Yes, you are right. One JVM instance is embed per Nginx Worker process. The number of Nginx Workers is generally the same with the number of CPU. If one Worker crashs the Nginx Master will create a new one so you don't worry about JVM crashs accidentally. Although there will be several JVM instances, there 's only one main thread attached with the Nginx Woker process. So the JVM instance uses less memory and no thread context switch cost in every JVM instance. In some cases If you can use only one JVM instance, you can set the Nginx Worker number to be 1 and set jvm_workers 1, nginx-clojure will create a thread pool with fixed number of thread. to handle requests for you. On Tuesday, January 14, 2014 5:50:34 PM UTC+8, Feng Shen wrote: Hi, Thanks for your work on nginx-clojure. It looks great! As I know Nginx spawns many processes(correct me if I am wrong), does that mean, there will be many JVM process? On Tuesday, January 14, 2014 4:44:18 PM UTC+8, Xfeep Zhang wrote: I have done the first one. The result is HEREhttps://github.com/ptaoussanis/clojure-web-server-benchmarks( https://github.com/ptaoussanis/clojure-web-server-benchmarks ) Thanks Taoussanis for his invitation to the project clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarkshosted on Github. On Tuesday, January 14, 2014 10:31:03 AM UTC+8, Xfeep Zhang wrote: You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzlehttps://launchpad.net/drizzlea no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.com wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Hi Mingli, Thanks for your suggestion. Nginx-Clojure is quite different from uwsgi when supports JVM. Nginx-Clojure make JVM embed into Nginx worker process. JVM and Nginx worker have the same memory process space. Nginx-Clojure heavy uses pointer operation just like C to handle memory with Nginx worker. If I'm not wrong, uwsgi create every process for every request, or shared JVM processes between request? When using Nginx-Clojure there's no IPC cost even no thread switch cost if jvm_workers = 0 which is default. So it's why Nginx-Clojure is so fast! On Wednesday, January 15, 2014 3:07:13 AM UTC+8, Mingli Yuan wrote: Hi, Xfeep, Thanks for your contribution, and the project looks interesting. For me, the idea of driving ring webapp behind nginx is not new. We use uwsgi to drive our ring app behind nginx in our production. uwsgi support JVM and ring for almost one year, and I think the code is relative stable right now. - it support a native protocol between nginx and uwsgi which is more efficient than http - it support unix socket - and a rich uwsgi api layer to provide some means to communicate between webapps - and according to the performance tests by the author, it is a little bit faster than jetty. It is on our production for half a year, quite stable, and very harmonious with the python app. I am not want to sale the solution of uwsgi, but it worth taking a look and make some comparison. Regards, Mingli On Tue, Jan 14, 2014 at 9:12 PM, Xfeep Zhang easy...@163.comjavascript: wrote: You are welcome! Yes, you are right. One JVM instance is embed per Nginx Worker process. The number of Nginx Workers is generally the same with the number of CPU. If one Worker crashs the Nginx Master will create a new one so you don't worry about JVM crashs accidentally. Although there will be several JVM instances, there 's only one main thread attached with the Nginx Woker process. So the JVM instance uses less memory and no thread context switch cost in every JVM instance. In some cases If you can use only one JVM instance, you can set the Nginx Worker number to be 1 and set jvm_workers 1, nginx-clojure will create a thread pool with fixed number of thread. to handle requests for you. On Tuesday, January 14, 2014 5:50:34 PM UTC+8, Feng Shen wrote: Hi, Thanks for your work on nginx-clojure. It looks great! As I know Nginx spawns many processes(correct me if I am wrong), does that mean, there will be many JVM process? On Tuesday, January 14, 2014 4:44:18 PM UTC+8, Xfeep Zhang wrote: I have done the first one. The result is HEREhttps://github.com/ptaoussanis/clojure-web-server-benchmarks( https://github.com/ptaoussanis/clojure-web-server-benchmarks ) Thanks Taoussanis for his invitation to the project clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarkshosted on Github. On Tuesday, January 14, 2014 10:31:03 AM UTC+8, Xfeep Zhang wrote: You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzle https://launchpad.net/drizzle a no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.com wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
I have check the uwsgi document again and the JVM integration document is HERE http://uwsgi-docs.readthedocs.org/en/latest/JWSGI.htmlhttp://uwsgi-docs.readthedocs.org/en/latest/JWSGI.html . When using uwsgi to integrate JVM, the JVM process is not the same process of Nginx worker. So it can not avoid IPC cost or socket cost ! So it will use more system handle (at least double ones) and more copy operataion between process and maybe more memory cost. So I think using uwsgi to integrate JVM maybe will not be so fast! On Wednesday, January 15, 2014 10:10:40 AM UTC+8, Xfeep Zhang wrote: Hi Mingli, Thanks for your suggestion. Nginx-Clojure is quite different from uwsgi when supports JVM. Nginx-Clojure make JVM embed into Nginx worker process. JVM and Nginx worker have the same memory process space. Nginx-Clojure heavy uses pointer operation just like C to handle memory with Nginx worker. If I'm not wrong, uwsgi create every process for every request, or shared JVM processes between request? When using Nginx-Clojure there's no IPC cost even no thread switch cost if jvm_workers = 0 which is default. So it's why Nginx-Clojure is so fast! On Wednesday, January 15, 2014 3:07:13 AM UTC+8, Mingli Yuan wrote: Hi, Xfeep, Thanks for your contribution, and the project looks interesting. For me, the idea of driving ring webapp behind nginx is not new. We use uwsgi to drive our ring app behind nginx in our production. uwsgi support JVM and ring for almost one year, and I think the code is relative stable right now. - it support a native protocol between nginx and uwsgi which is more efficient than http - it support unix socket - and a rich uwsgi api layer to provide some means to communicate between webapps - and according to the performance tests by the author, it is a little bit faster than jetty. It is on our production for half a year, quite stable, and very harmonious with the python app. I am not want to sale the solution of uwsgi, but it worth taking a look and make some comparison. Regards, Mingli On Tue, Jan 14, 2014 at 9:12 PM, Xfeep Zhang easy...@163.com wrote: You are welcome! Yes, you are right. One JVM instance is embed per Nginx Worker process. The number of Nginx Workers is generally the same with the number of CPU. If one Worker crashs the Nginx Master will create a new one so you don't worry about JVM crashs accidentally. Although there will be several JVM instances, there 's only one main thread attached with the Nginx Woker process. So the JVM instance uses less memory and no thread context switch cost in every JVM instance. In some cases If you can use only one JVM instance, you can set the Nginx Worker number to be 1 and set jvm_workers 1, nginx-clojure will create a thread pool with fixed number of thread. to handle requests for you. On Tuesday, January 14, 2014 5:50:34 PM UTC+8, Feng Shen wrote: Hi, Thanks for your work on nginx-clojure. It looks great! As I know Nginx spawns many processes(correct me if I am wrong), does that mean, there will be many JVM process? On Tuesday, January 14, 2014 4:44:18 PM UTC+8, Xfeep Zhang wrote: I have done the first one. The result is HEREhttps://github.com/ptaoussanis/clojure-web-server-benchmarks( https://github.com/ptaoussanis/clojure-web-server-benchmarks ) Thanks Taoussanis for his invitation to the project clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarkshosted on Github. On Tuesday, January 14, 2014 10:31:03 AM UTC+8, Xfeep Zhang wrote: You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzle https://launchpad.net/drizzle a no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.comwrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Hi Mingli, Thanks for your suggestion. Nginx-Clojure is quite different from uwsgi when supports JVM. Nginx-Clojure make JVM embed into Nginx worker process. JVM and Nginx worker have the same memory process space. Nginx-Clojure heavy uses pointer operation just like C to handle memory with Nginx worker. If I'm not wrong, uwsgi create every process for every request, or shared JVM processes between request? uWSGI creates a pool of processes with a number of threads in each and then they are used for the whole server life cycle. (a pretty standard behaviour) When using Nginx-Clojure there's no IPC cost even no thread switch cost if jvm_workers = 0 which is default. So it's why Nginx-Clojure is so fast! i strongly suggest you to avoid the performance as a selling point, your project is cool but not for performance (and you are using a pipe to transfer requests data from nginx to the jvm so there ipc in place too, even if you can improve things using OS-specific syscall like splice). It is cool because it simplify deployments, but nginx is not an application server, so things like multi-strategy gracefully reloads, stuck-requests managers, offloading and so on are not available. Regarding the ipc problem, you generally put a load balancer on front of your app, so there is always ipc Do not get me wrong, as i have already said your project is cool, but you should focus not on performance but on better integration with the nginx api for writing non-blocking apps (if possible), something at uWSGI unfortunately failed (the JVM has no support for switching stacks out of the box, something that is required for async mode in uWSGI) -- Roberto De Ioris http://unbit.it -- -- 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.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easyj...@163.com wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request) (2) http-kit 2.1.16 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.691 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1340 bytes HTML transferred: 150 bytes Requests per second:37165.27 [#/sec] (mean) Time per request: 269.068 [ms] (mean) Time per request: 0.027 [ms] (mean, across all concurrent requests) Transfer rate: 4863.42 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 72 118 46.21141094 Processing:31 134 26.1136 344 Waiting: 21 81 33.5 71 273 Total:183 252 43.82511435 Percentage of the requests served within a certain time (ms) 50%251 66%258 75%259 80%261 90%263 95%263 98%265 99%266 100% 1435 (longest request) (3) ring-jetty Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 9.740 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1670 bytes HTML transferred: 150 bytes Requests per second:10267.16 [#/sec] (mean) Time per request: 973.979 [ms] (mean) Time per request: 0.097 [ms] (mean, across all concurrent requests)
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
You're welcome. I think there are several difficult phases : (1) update the test program in clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarks, make the some packages to be the latest. (eg. http-kit from 1.3.0-alpha2 -- 2.1.16) and add nginx-php testing (2) test about real world size contents by group eg. tiny, small, medium, huge. (3) test about real world connection circumstances where a lot of connection is inactive but keep open. (4) try some real asynchronous test to fetch external resources (eg. rest service , db) before response to the client. eg. using libdrizzlehttps://launchpad.net/drizzlea no-blocking mysql client from https://launchpad.net/drizzle On Tuesday, January 14, 2014 2:41:50 AM UTC+8, Sergey Didenko wrote: Looks very interesting, thank you for your work! I wonder how this is going to improve latency in comparison to nginx + http-kit for some real world test that is not using heavy DB operations. On Mon, Jan 13, 2014 at 5:57 AM, Xfeep Zhang easy...@163.comjavascript: wrote: So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request) (2) http-kit 2.1.16 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.691 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1340 bytes HTML transferred: 150 bytes Requests per second:37165.27 [#/sec] (mean) Time per request: 269.068 [ms] (mean) Time per request: 0.027 [ms] (mean, across all concurrent requests) Transfer rate: 4863.42 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 72 118 46.21141094
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Sorry for my mistake! 1. In the static file test, the result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes Concurrency Level: 1 Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes Concurrency Level: 1 Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request) (2) http-kit 2.1.16 Document Path: / Document Length:15 bytes Concurrency Level: 1 Time taken for tests: 2.691 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1340 bytes HTML transferred: 150 bytes Requests per second:37165.27 [#/sec] (mean) Time per request: 269.068 [ms] (mean) Time per request: 0.027 [ms] (mean, across all concurrent requests) Transfer rate: 4863.42 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 72 118 46.21141094 Processing:31 134 26.1136 344 Waiting: 21 81 33.5 71 273 Total:183 252 43.82511435 Percentage of the requests served within a certain time (ms) 50%251 66%258 75%259 80%261 90%263 95%263 98%265 99%266 100% 1435 (longest request) (3) ring-jetty Document Path: / Document Length:15 bytes Concurrency Level: 1 Time taken for tests: 9.740 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1670 bytes HTML transferred: 150 bytes Requests per second:10267.16 [#/sec] (mean) Time per request: 973.979 [ms] (mean) Time per request: 0.097 [ms] (mean, across all concurrent requests) Transfer rate: 1674.43 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect:0 193 399.8 113006 Processing: 0 51 207.6 57050 Waiting:0 39 204.5 47050 Total: 0 244 482.0 288080 Percentage of the requests served within a certain time (ms) 50% 28 66% 79 75%283 80%306 90% 1009 95% 1067 98% 1283 99% 1886 100% 8080 (longest request) On Friday, January 10, 2014 8:23:02 AM UTC+8, Julien wrote: Impressive! Did you run some benchmark? How does it compare to ring-jetty and http-kit? Julien Le jeudi 9 janvier 2014 12:42:31 UTC-3, Xfeep Zhang a écrit : [image: Alt text]https://github.com/xfeep/nginx-clojure/blob/master/logo.pngNginx-Clojure is a Nginx http://nginx.org/ module for embedding Clojure or Java programs, typically those Ringhttps://github.com/ring-clojure/ring/blob/master/SPECbased handlers. There are some core features : 1.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request) (2) http-kit 2.1.16 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.691 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1340 bytes HTML transferred: 150 bytes Requests per second:37165.27 [#/sec] (mean) Time per request: 269.068 [ms] (mean) Time per request: 0.027 [ms] (mean, across all concurrent requests) Transfer rate: 4863.42 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 72 118 46.21141094 Processing:31 134 26.1136 344 Waiting: 21 81 33.5 71 273 Total:183 252 43.82511435 Percentage of the requests served within a certain time (ms) 50%251 66%258 75%259 80%261 90%263 95%263 98%265 99%266 100% 1435 (longest request) (3) ring-jetty Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 9.740 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1670 bytes HTML transferred: 150 bytes Requests per second:10267.16 [#/sec] (mean) Time per request: 973.979 [ms] (mean) Time per request: 0.097 [ms] (mean, across all concurrent requests) Transfer rate: 1674.43 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect:0 193 399.8 113006 Processing: 0 51 207.6 57050 Waiting:0 39 204.5 47050 Total: 0 244 482.0 288080 Percentage of the requests served within a certain time (ms) 50% 28 66% 79 75%283 80%306 90% 1009 95% 1067 98% 1283 99% 1886 100% 8080 (longest request) On Friday, January 10, 2014 8:23:02 AM UTC+8, Julien wrote: Impressive! Did you run some benchmark? How does it compare to ring-jetty and http-kit? Julien Le jeudi 9 janvier 2014 12:42:31 UTC-3, Xfeep Zhang a écrit : [image: Alt text]https://github.com/xfeep/nginx-clojure/blob/master/logo.pngNginx-Clojure is a Nginx http://nginx.org/ module for embedding Clojure or Java programs, typically those Ringhttps://github.com/ring-clojure/ring/blob/master/SPECbased handlers. There are some core
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
So far I have found why nginx-clojure is slower than http-kit when 1 concurrents. (when = 1000 concurrents nginx-clojure is faster than http-kit.) I have set too many connections per nginx worker (worker_connections = 2) . This make nginx only use one worker to handle ab requests (every request is tiny). I plan to take note of c-erlang-java-performancehttp://timyang.net/programming/c-erlang-java-performance/and fork clojure-web-server-benchmarkshttps://github.com/ptaoussanis/clojure-web-server-benchmarksto do some real world tests. On Sunday, January 12, 2014 11:21:06 PM UTC+8, Xfeep Zhang wrote: Sorry for my mistake! 1. In the static file test, the ring-jetty result is about 10 concurrents. NOT 1 concurrents (Concurrency Level: 10 in the ab report ). 2. In the small string test, All results about three server are about 10 concurrents. NOT 1 concurrents. There are right results about these two mistake : 1. static file test (3) ring-jetty more bad than 10 concurrents === Document Path: / Document Length:29686 bytes *Concurrency Level: 1* Time taken for tests: 6.303 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 298220 bytes HTML transferred: 296860 bytes Requests per second:15864.43 [#/sec] (mean) Time per request: 630.341 [ms] (mean) Time per request: 0.063 [ms] (mean, across all concurrent requests) Transfer rate: 462020.65 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 12 328 535.0 433041 Processing:25 124 112.9 963523 Waiting:8 47 99.4 283523 Total: 52 452 544.51574546 Percentage of the requests served within a certain time (ms) 50%157 66%305 75% 1071 80% 1102 90% 1139 95% 1155 98% 1462 99% 3100 100% 4546 (longest request) 2. simple string (1 concurrents) http-kit is the fastest. But nginx-clojure is too young and has vast room for growth ：） (1) nginx-clojure-0.1.0 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.834 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1700 bytes HTML transferred: 150 bytes Requests per second:35291.16 [#/sec] (mean) Time per request: 283.357 [ms] (mean) Time per request: 0.028 [ms] (mean, across all concurrent requests) Transfer rate: 5858.88 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 51 118 21.6118 178 Processing:73 150 33.8146 263 Waiting: 42 110 32.0104 246 Total:177 268 25.6269 327 Percentage of the requests served within a certain time (ms) 50%269 66%278 75%285 80%288 90%297 95%309 98%314 99%318 100%327 (longest request) (2) http-kit 2.1.16 Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 2.691 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1340 bytes HTML transferred: 150 bytes Requests per second:37165.27 [#/sec] (mean) Time per request: 269.068 [ms] (mean) Time per request: 0.027 [ms] (mean, across all concurrent requests) Transfer rate: 4863.42 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect: 72 118 46.21141094 Processing:31 134 26.1136 344 Waiting: 21 81 33.5 71 273 Total:183 252 43.82511435 Percentage of the requests served within a certain time (ms) 50%251 66%258 75%259 80%261 90%263 95%263 98%265 99%266 100% 1435 (longest request) (3) ring-jetty Document Path: / Document Length:15 bytes *Concurrency Level: 1* Time taken for tests: 9.740 seconds Complete requests: 10 Failed requests:0 Write errors: 0 Total transferred: 1670 bytes HTML transferred: 150 bytes Requests per second:10267.16 [#/sec] (mean) Time per request: 973.979 [ms] (mean) Time per request: 0.097 [ms] (mean, across all concurrent requests) Transfer rate: 1674.43 [Kbytes/sec] received Connection Times (ms) min mean[+/-sd] median max Connect:0 193 399.8 113006 Processing: 0 51 207.6 57050 Waiting:0 39 204.5 47050 Total:
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Thank you! I think it's useful.
I have done some simple tests. But I think general performance test may be
meaningless regardless of real world requirements.
os : ubuntu 13.10 64bit
memory: 16G
cpu: intel i7 4700MQ (4 cores 2.4GHz)
1. static file test
file: 29.7k (real contents from https://groups.drupal.org/node/167984)
ring handler :
(def test-handler [req]
{:status 200
:headers {content-type text/html}
:body (java.io.File. resources/index.html) })
warmed by ab -n 40 -c 1 http://localhost:${port}/;
test command:
ab -n 10 -c 1 http://localhost:${port}/
(1) nginx-clojure-0.1.0
===
Document Path: /
Document Length:29686 bytes
Concurrency Level: 1
Time taken for tests: 3.464 seconds
Complete requests: 10
Failed requests:0
Write errors: 0
Total transferred: 299120 bytes
HTML transferred: 296860 bytes
Requests per second:28867.39 [#/sec] (mean)
Time per request: 346.412 [ms] (mean)
Time per request: 0.035 [ms] (mean, across all concurrent requests)
Transfer rate: 843243.63 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 18 113 208.6 743082
Processing:82 219 51.6225 748
Waiting: 13 78 37.6 70 604
Total:102 332 216.93103190
Percentage of the requests served within a certain time (ms)
50%310
66%329
75%337
80%341
90%348
95%423
98% 1295
99% 1309
100% 3190 (longest request)
(2) http-kit 2.1.16
===
Document Path: /
Document Length:29686 bytes
Concurrency Level: 1
Time taken for tests: 4.104 seconds
Complete requests: 10
Failed requests:0
Write errors: 0
Total transferred: 298080 bytes
HTML transferred: 296860 bytes
Requests per second:24363.92 [#/sec] (mean)
Time per request: 410.443 [ms] (mean)
Time per request: 0.041 [ms] (mean, across all concurrent requests)
Transfer rate: 709218.63 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect: 49 130 229.3 983062
Processing: 124 269 38.6267 603
Waiting: 39 87 24.0 88 353
Total:243 398 233.73693665
Percentage of the requests served within a certain time (ms)
50%369
66%379
75%387
80%395
90%415
95%443
98% 1310
99% 1396
100% 3665 (longest request)
(3) ring-jetty
===
Document Path: /
Document Length:29686 bytes
Concurrency Level: 10
Time taken for tests: 4.991 seconds
Complete requests: 10
Failed requests:0
Write errors: 0
Total transferred: 298220 bytes
HTML transferred: 296860 bytes
Requests per second:20037.89 [#/sec] (mean)
Time per request: 0.499 [ms] (mean)
Time per request: 0.050 [ms] (mean, across all concurrent requests)
Transfer rate: 583564.46 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect:00 0.0 0 1
Processing: 00 0.8 0 75
Waiting:00 0.8 0 75
Total: 00 0.8 0 75
Percentage of the requests served within a certain time (ms)
50% 0
66% 1
75% 1
80% 1
90% 1
95% 1
98% 1
99% 1
100% 75 (longest request)
2. simple string
warmed by ab -n 40 -c 1 http://localhost:${port}/;
ring handler :
(def test-handler [req]
{:status 200
:headers {content-type text/html}
:body Hello, Clojure! })
test command:
ab -n 10 -c 1 http://localhost:${port}/
(1) nginx-clojure-0.1.0
==
Document Path: /
Document Length:15 bytes
Concurrency Level: 10
Time taken for tests: 1.952 seconds
Complete requests: 10
Failed requests:0
Write errors: 0
Total transferred: 1700 bytes
HTML transferred: 150 bytes
Requests per second:51241.03 [#/sec] (mean)
Time per request: 0.195 [ms] (mean)
Time per request: 0.020 [ms] (mean, across all concurrent requests)
Transfer rate: 8506.81 [Kbytes/sec] received
Connection Times (ms)
min mean[+/-sd] median max
Connect:00 0.0 0 1
Processing: 00 0.1 0 3
Waiting:00 0.1 0 3
Total: 00 0.1 0 4
Percentage of the requests served within a certain time (ms)
50% 0
66% 0
75% 0
80% 0
90% 0
Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
[image: Alt text]https://github.com/xfeep/nginx-clojure/blob/master/logo.pngNginx-Clojure is a Nginx http://nginx.org/ module for embedding Clojure or Java programs, typically those Ringhttps://github.com/ring-clojure/ring/blob/master/SPECbased handlers. There are some core features : 1. Compatible with Ringhttps://github.com/ring-clojure/ring/blob/master/SPECand obviously supports those Ring based frameworks, such as Compojure etc. 2. One of benifits of Nginx http://nginx.org/ is worker processes are automatically restarted by a master process if they crash 3. Utilizes lazy headers and direct memory operation between Nginxhttp://nginx.org/and JVM to fast handle dynamic contents from Clojure or Java code. 4. Utilizes Nginx http://nginx.org/ zero copy file sending mechanism to fast handle static contents controlled by Clojure or Java code. 5. Supports Linux x64, Win32 and Mac OS X Use Nginx-Clojure, you can deploy clojure web app on Nginx without any Java web server. For more detials please check Nginx-Clojure github sitehttps://github.com/xfeep/nginx-clojure. -- -- 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.
Re: Nginx-Clojure Let You Deploy Clojure Web App on Nginx Without Any Java Web Server
Impressive! Did you run some benchmark? How does it compare to ring-jetty and http-kit? Julien Le jeudi 9 janvier 2014 12:42:31 UTC-3, Xfeep Zhang a écrit : [image: Alt text]https://github.com/xfeep/nginx-clojure/blob/master/logo.pngNginx-Clojure is a Nginx http://nginx.org/ module for embedding Clojure or Java programs, typically those Ringhttps://github.com/ring-clojure/ring/blob/master/SPECbased handlers. There are some core features : 1. Compatible with Ringhttps://github.com/ring-clojure/ring/blob/master/SPECand obviously supports those Ring based frameworks, such as Compojure etc. 2. One of benifits of Nginx http://nginx.org/ is worker processes are automatically restarted by a master process if they crash 3. Utilizes lazy headers and direct memory operation between Nginxhttp://nginx.org/and JVM to fast handle dynamic contents from Clojure or Java code. 4. Utilizes Nginx http://nginx.org/ zero copy file sending mechanism to fast handle static contents controlled by Clojure or Java code. 5. Supports Linux x64, Win32 and Mac OS X Use Nginx-Clojure, you can deploy clojure web app on Nginx without any Java web server. For more detials please check Nginx-Clojure github sitehttps://github.com/xfeep/nginx-clojure. -- -- 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.
