New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
There ya go bragging again :P, and rightly so, centminmod performs great.
What custom_config.inc settings do you use?
For those benchmark comparisons ? no persistent config file custom_config.inc settings - it's just out of box centmin mod 123.09beta01 defaults.
How many VPS's and how to do you find managing multiple Wordpress sites on 1 VPS, what tools do you use to do this, etc. Thanks!
SmushIT only allow lossless image optimization
With LiteSpeed Cache's Image Optimization you can use lossy optimization and also get .webp of the images
The "Litespeed" limitation is only on the caching functionality, and in this case anything that requires .htaccess, so you can actually use the their free image optimization with centminmod.
Although I'm having issues on actually making WordPress use said .webp images as LiteSpeed only optimized the resized images
Yeah I know you can use LSCache image optimization. Though there's other wp plugins and methods too https://community.centminmod.com/threads/playing-with-ewww-image-optimizer-wordpress-plugin.12573/ and with webp
EWWW Image Optimizer webp conversion is ok - output image stats for wp images using my optimise-images.sh profiling mode
But compared to my optimise-images.sh script https://github.com/centminmod/optimise-images + webp conversion posted here for wordpress older tests
Currently have each of my web design clients on a VPS each. My personal sites are spread over 3 VPS's, 20+/- lightsail or aws instances and a couple reseller accounts. But as far as traditional VPS'a are concerned I dont have many. Only 3 foe websites and 2 for backups. Not sure what to classify lightsail and aws as.
should be able to use a simple apache .htaccess rewrite rule to tell litespeed to serve .webp version if it exists just like nginx does for webp. I did write up for centmin mod nginx webp at https://centminmod.com/webp/
example of batch converting images to webp and auto generating nginx rule to use for webp serving at https://github.com/centminmod/optimise-images/blob/master/examples/examples-optimise-webp-nginx-300417.md
edit: lscache -> advanced optimisations -> media -> Image WebP Replacement = enable will do the apache .htaccess rewrite it seems
I'm running centminmod, the issue I'm having is Lscache not optimizing the main image, only the wordpress resized versions, but it doesn't matter. This was just a test blog
i see.. there's option in media advance settings -> Optimize Original Images
Did you check with apache bench or other stress testing software, for real life scenario?
As Cache Enable is going to put more stress on PHP when there is high load on the server.
Apart from that LSCache (Server side module) is very intelligent, and when you combine it with LSCache plugin it becomes a great combination. Because LSCache plugin handle cache invalidation intelligently, you need not worry about cache expiration time or invalid content as it use tag-based invalidation and talks directly to LSCache module on OpenLiteSpeed/LSWS so there is no proxy/middle man involved.
OpenLiteSpeed is not going to fork single PHP instance, unless there is a invalid cache copy.
Would love for you to run some tests on OpenLiteSpeed or even include it in centminmod for command line enthusiasts.
Yes there's plans for further testing and openlitespeed is on to do list for integration into Centmin Mod outlined at https://github.com/centminmod/centminmod/projects/1
Though current focus is on pagespeed as I am testing out my new gitools.sh script which queries Google PageSpeed Insights, Webpagetest and GTMetrix APIs to run page speed tests and sends metrics to a Slack channel you can define https://github.com/centminmod/google-insights-api-tools
Yes i am obsessed with server performance but also addicted to optimisations for page load speeds - particularly perceived visual render times
That is great, show some love to the folks at LiteSpeed. Because with your optimization skills it will get more better.
Visual rendering will be more better with LSCache on real world scenarios when there is more stress on server (particularly PHP side). Plus invalidation also matters as Cache Enabler invalidation is not very effective.
Some options on LSCache plugin are also disabled to avoid any conflicts, but end users can tweak and test according to their environment and have even better visual rendering.
For CyberPanel side you now have option to either install 1 or 2 instance version more details at: https://forums.cyberpanel.net/discussion/459/cyberpanel-1-7-rc-email-limits-spamassassin-cli-rewrite-context-email-forwarding#latest
You are of course 100% correct. And you talk from experience. But this is not how most people think. Normal thinking is that this one new WP site will change the world forever and will have millions of hits in no time.
Doing CyberPanel 1.7 RC tests now too - great job on it's development so far !
Round 2 testing including HTTP/2 HTTPS benchmarks and WPT mobile and desktop with CyberPanel 1.7 RC vs Centmin Mod 123.09beta01
You can jump to the HTTP/2 HTTPS benchmarks here. Bare in mind when looking at those results, Centmin Mod Nginx out of the box is between 40-280% faster than other Nginx web servers (with optional another 20-40% you can optionally squeeze out of it with advanced configurations). There's potentially also upto another 60% boost when pairing GCC 8.2 with Intel Skylake detected cpus (GCC 8.x support is work in progress for Centmin Mod).
You are doing the lord's work for sure.
I found your optimize images shell script - nice work but can you point me to a quick way to use ImageMagick to batch optimize images? No matter what I do, google pagespeed says the images aren't small enough and can be compressed further. I'm sure this is a problem a lot of people have.
Great work and thanks for making it all free. Just amazing...
@eva2000
Thanks for the feedback.
The difference you are seeing in results are due to page size.
Setup using CMM option 22 (default with no other optimizations):
Where was on your test page size is:
For single run, and when it is tested concurrently difference in data size becomes significant.
When I tested using -c500 -n10000 $domain
I got following on OLS
h2load -t1 -H 'Accept-Encoding: gzip' -c500 -n10000 $domain starting benchmark... spawning thread #0: 500 total client(s). 10000 total requests TLS Protocol: TLSv1.2 Cipher: ECDHE-RSA-AES128-GCM-SHA256 Server Temp Key: ECDH P-256 256 bits Application protocol: h2 progress: 10% done progress: 20% done progress: 30% done progress: 40% done progress: 50% done progress: 60% done progress: 70% done progress: 80% done progress: 90% done progress: 100% done finished in 2.93s, 3414.54 req/s, 36.20MB/s requests: 10000 total, 10000 started, 10000 done, 10000 succeeded, 0 failed, 0 errored, 0 timeout status codes: 10000 2xx, 0 3xx, 0 4xx, 0 5xx traffic: 106.01MB (111155039) total, 313.03KB (320539) headers (space savings 94.12%), 105.33MB (110450000) data min max mean sd +/- sd time for request: 18.86ms 230.23ms 103.33ms 39.48ms 84.19% time for connect: 217.40ms 1.30s 662.60ms 379.27ms 51.60% time to 1st byte: 255.13ms 1.42s 815.61ms 350.64ms 51.60% req/s : 6.84 8.09 7.35 0.36 51.60%on CMM:
h2load -t1 -H 'Accept-Encoding: gzip' -c500 -n10000 $domain starting benchmark... spawning thread #0: 500 total client(s). 10000 total requests TLS Protocol: TLSv1.2 Cipher: ECDHE-RSA-AES128-GCM-SHA256 Server Temp Key: ECDH P-256 256 bits Application protocol: h2 progress: 10% done progress: 20% done progress: 30% done progress: 40% done progress: 50% done progress: 60% done progress: 70% done progress: 80% done progress: 90% done progress: 100% done finished in 9.64s, 1037.50 req/s, 11.13MB/s requests: 10000 total, 10000 started, 10000 done, 10000 succeeded, 0 failed, 0 errored, 0 timeout status codes: 10000 2xx, 0 3xx, 0 4xx, 0 5xx traffic: 107.23MB (112441567) total, 2.92MB (3057067) headers (space savings 23.57%), 104.04MB (109090000) data min max mean sd +/- sd time for request: 2.29ms 217.67ms 64.67ms 63.33ms 67.02% time for connect: 261.55ms 9.41s 4.17s 3.15s 47.00% time to 1st byte: 303.09ms 9.43s 4.22s 3.11s 46.80% req/s : 2.08 6.20 4.24 1.53 43.20%PS: I am also not getting Cache Enabler hits when on SSL (CMM), might be doing something wrong with setup?
Have you made any further plugin side optimization to reduce the page size?
Thank you for your time.
tests were done with respective wp plugin cached settings enabled https://community.centminmod.com/threads/wordpress-webpagetest-pagespeed-comparison-for-cyberpanel-openlitespeed-vs-centmin-mod-lemp.15144/#post-64938 so for centmin mod site with cache enabler selected from centmin.sh menu option 22, + autooptimize, clearfy and smushit so probably were the page size came down from
tests are comparison their respective wordpress configured cache/optimisation plugin settings
if i disabled autoptimize and clearfy i get this for centmin mod + cache enabler
page size is up but performance isn't that drastically different so could be cache enabler not caching for you though even 1000 req/s isn't non-cached as php can't perform like that by itself
renable autoptimize + clearfy and get this
You can use Nginx as a reverse proxy with Apache.
@cyberpersons you just gave me an idea to improve my wp setups - hack cache enabler to support zopfli compression instead of gzip - brings index pre-compressed size down from 6.9KB to 6.7KB !
not php coder so not 100% sure but seems to work
Great work.
I believe amount of data transferred is issue here. With out of the box CMM setup, I got:
I am not sure how you are getting this with out of box CMM setup:
Even with that much difference OpenLiteSpeed performed well. But again it brings us down to intelligent cache, LiteSpeed Cache plugin have very intelligent cache invalidation.
You must apply some of your optimization skills to OpenLiteSpeed, you will love it.
Feel sorry for OP for getting his thread off-topic.
the reduction is due to enabled autoptimize + clearfy settings - i tested without autoptimize and only clearfy and size went up but still speed was better with lower TTFB with just cache enabler + clearfy https://community.centminmod.com/posts/65160/
but zopfli hack was a no go - worked in my real chrome/opera browsers but webpagetest gave a decoding error so guess my hack didn't work heh
yeah OP sorry - we got carried away ^_^
Did you use the betainstaller72 ?
yes but defaulting to PHP 71
strange then - using some code in betainstaller72.sh to on the fly grab latest php 7.2 version from php download page unless it grabbed wrong one for you (there's a grep 7.2 in there so hard to get 7.1 heh)
nginx + php-fpm definitely. You can combine it with nginx micro cache. Here is a good guide. You will get super fast server and low on resources.
I orgasmed thrice just reading @cyberpersons and @eva2000's exchanges
now back to topic...
LOL well prepare yourself again.. Autoptimize wp plugin has hidden API support for pre-gzip compressed css/js files which is just perfect if like Centmin Mod Nginx, your Nginx server has gzip_static directive enabled to serve pre-gzipped files which can be up to 10x times faster in terms requests/s handled than on the fly gzip compressed files
I created my first ever Wordpress plugin at Autoptimize Gzip to enable Autoptimize pre-gzip support via an API hook and did some benchmarks https://community.centminmod.com/posts/65227/
So might be useful for anyone using Autoptimize wp plugin
HHVM-proxygen 8)
OpenLiteSpeed by default compress static resources and won't do again until they are modified.
Plus there will be another change coming to OpenLiteSpeed that combines small writes to a bigger one giving almost 25% performance boost.