Zoinkies Thanks for the reply, have had bad experience with OVH in the past so we'd like to avoid it. Colo would surely be the best option, but it will limit us due to fund requirement. We will not be able to quickly scale to multiple locations. Thinking of going with Ryzen 7950X chips and I think they should give satisfactory performance. But the question still remains, why is the performance that low with Enhance? I tested with both Apache and OLS here and compared with Runcloud.

      dynamo Yea, avoid OVH, I had a 1-1 relationship with them, and they still caused me major downtime. If you're interested, I am an investor in multi web companies, if your marketing is good and current customer base and growth it decent I'd happily invest. I personally, if you're looking for absolute performance go the intel route, but if you're looking for okay performance, AMD. I think Runcloud has optimised PHP versions, and so does Kinsta etc.

          7 days later

          Hi everyone, this is my very first post as I just bought an Enhance license (btw, I thought there were 30 days free trial, but I had to pay the minimum 5 USD... anyway that's not the problem, it won't break the bank.)

          The real problem I encountered is abysmal performance from Enhance compared to Runcloud. I'm a Runcloud customer since a roughly year, but I like to also check other solutions...

          I needed to deploy a new server to take some load off my other older servers, so considering I had a nice weekend in front of me, I wanted to finally try Enhance and this is what I did:

          • I got a nice and fresh dedicated EX44 from Hetzner (i5-13500, 64GB, 2x512GB Nvme) with Ubuntu 22.04 LTS
          • I immediately installed Enhance on it, nothing else.
          • Once I had enhance running, I deployed an empty WP with Nginx.
          • I installed the plugin WPPerformanceTester and nothing else, and ran it.

          And... I was shocked by the results, I ran it multiple times, and even tried to switch Nginx to Apache and to OLS.

          • The average result I got for Server Performance Benchmark was around 13sec. (quite bad considering the server it was running on)
          • But the ugly part was the Wordpress Performance Benchmark where, no matter what I tried, never got over 150 queries/sec! Always between 120-150 q/sec... ugh...

          Considering that I'm getting over 1600 q/sec. on a crowded, standard 3 cores VPS from Hetzner, something seems wrong.

          So I reset the server and made a fresh OS install (still Ubuntu 22.04 LTS) and this time installed my usual Runcloud with Nginx.
          Same story as above: installed an empty WP and the same performance tester plugin.
          Server Performance Benchmark: around 5.5 sec. (more than 100% increase in performance)
          Wordpress Performance Benchmark: roughly 5500 queries/sec. (an average of 45 times more performance)

          So for the moment I kept my Runcloud install and already moved some heavy woocommerce websites from my old servers.

          Now, my question: are these normal performances on Enhance? Or what could be wrong considering that everything was freshly installed, and it's a dedicated server (so obviously no noisy neighbours, etc) ?

          Too bad, I was curious to try Enhance but the awful perfs compared to Runcloud for the moment say it's a no-no...

          Any advice?

          Since everyone's workload and server spec is different, Enhance doesn't make any performance optimisations "out of the box". It gives you the tools you need (resource management, server migrations, web servers with full page caching, my.cnf/php.ini templates, etc).

          Regarding the WordPress performance test, there could be many factors but one possibility is that the Runcloud server is running MariaDB. MariaDB is a fork of MySQL 5.7 and therefore still has the query cache. Enhance deploys MySQL 8.0 - the MySQL team decided to remove the query cache for version 8. If the query in the benchmark is benefiting from the query cache then that might explain the huge difference in results.

          In the real world, you're unlikely to see the same behaviour since the table is likely to be changing - every time it changes the query cache is invalidated.

          Additionally on a real live WordPress site you would almost certainly be using full page caching and the database would not be hit.

          Enhance will support MariaDB in September for customers who want to use it. There are definitely upsides and downsides.

            15 days later

            Adam Since everyone's workload and server spec is different, Enhance doesn't make any performance optimisations "out of the box". It gives you the tools you need (resource management, server migrations, web servers with full page caching, my.cnf/php.ini templates, etc).

            Considering "High Performance" is the 3rd feature title on the homepage, this surprises me. If this is a core belief of ECP I hope we at least get extensive official guides on how to perf opt every single part of the ECP stack.

            I have my eye on ECP specifically because I'm not a pro linux sysadmin, I won't be comfortable figuring this out on my own, and I expect there are lots of people like me. But if ECP is going to be slow for us, then I'll have to look elsewhere.

            Additionally on a real live WordPress site you would almost certainly be using full page caching and the database would not be hit.

            Sure, websites that can be cached should be cached, but we should also optimize for highly dynamic sites. Database requests, and all other parts of the chain. Right?

            • Adam replied to this.

                timbr Considering "High Performance" is the 3rd feature title on the homepage.

                It is but we then list the tools we make available so you can configure the environment to your server specs and usage case - shared hosting, managed hosting, self-hosting, etc. We don't make assumptions about your environment.

                In the future we can probably offer suggested configurations based on the server spec. For example we might detect how much RAM the server has and ask whether you intend to offer mass hosting and based on those answers we might recommend per-website cgroup limits or my.cnf optimisations. However this kind of functionality will be developed after all the outstanding core features.

                timbr I'm not a pro linux sysadmin,

                You don't need to be a pro sysadmin but we do assume a degree of Linux/hosting knowledge. In the very near future, customers needing more assistance will be able to purchase Enhance through managed hosting partners who will offer this level of support.

                    Anyone want to share their my.cnf settings? I'm getting 39 queries per second 😂😅

                      ss88us You can write to me on Discord - discordapp.com/users/923501960009744415
                      I will try to help.

                          AdamM I tried to find you but couldn't. Can you please ping me on discord badboybackagain#7564

                            Hey @AdamM,

                            Thank you for your offer, but I found out the culprit!

                            My MySQL instance had (this was a default setting)

                            innodb_flush_log_at_trx_commit = 1

                            Changing this to

                            innodb_flush_log_at_trx_commit = 2

                            It went from 39 queries per second to 3246 queries per second with a Post Benchmark of 9.5+ for 100 inserts. I did the same test with 1,000 and everything was less than 1 second and a 9.45+ score. I think I'm good!!

                            Edit: Updated innodb_flush_log_at_trx_commit = 2 after reading MySQL manual.

                              10 months later

                              This was a solid thread with a lot of great info.

                              I'm currently looking at dedicated servers and of course all I see are EPYCs and Ryzens; however, there's rarely info about specific models or overall computational power / threads.

                              With that being said, and since this is thread is a year or so old, what is everyone looking for now?

                              In no particular order,

                              EPYC 9634
                              Ryzen 9 3900X
                              Ryzen 9 5950X
                              Ryzen 9 7950X
                              Ryzen 7 7700
                              Ryzen 7 Pro 8750GE

                              What's your take?

                              Andreas PHP performance is mostly about single core performance. Capacity (ie concurrent requests & high traffic sites) is about number of cores. If you need the best performance go with the fastest Ryzen. If you need capacity go with EPYC.

                              Worth noting the Ryzen is a consumer CPU and a lot of providers don't use ECC memory. Some providers cut corners and don't have proper cooling or even professional server racks. Good dedicated providers using Ryzen do exist though. Best I've personally worked with is: https://ioflood.com/

                                4 months later
                                Follow @enhancecp