PC SOFT

FORUMS PROFESSIONNELS
WINDEVWEBDEV et WINDEV Mobile

Accueil → WINDEV 22 → WebDev Performance
WebDev Performance
Débuté par JP, 04 juin 2018 12:50 - 3 réponses
Posté le 04 juin 2018 - 12:50
Hi All

A question on performance and scalability for WebDev:

Q1) Can a WebDev website handle thousands of concurrent users?

Q2) In a scenario where one expects to handle thousands of users would WebDev be a good choice or not really?

Assume servers can be top end and one can add multiple servers to a farm.
Posté le 04 juin 2018 - 14:01
Hi JP

Sorry, but your question doesn't make sense...

Yes webdev can handle thousands of users if you configure/scale correctly your hardware... but that's the same than with ANY tool, and the scalabilité/hardware required will depend in big part on how optimized your code is...

Best regards
Posté le 04 juin 2018 - 20:44
Fabrice

Well of course development skills and hardware etc. are major influences but also different web platforms have different performance ratings. A WebDev server might be faster or slower than another server on the same hardware. Of course, such a comparison must assume both tools/sites are equally well developed by competent developers in their respective fields and using the same hardware setups. Such assumptions are implicit in such a comparison.

Similar example; an application developed by a competent C programmer vs. same application developed by a competent Visual Basic programmer on the same hardware ... C application will perform faster in probably all cases. It is just natively faster.

So for clarity - given the same server hardware (quantity and quality) and 2 sites developed in different platforms, one WebDev and one something else, by competent developers in each case, would WebDev be significantly slower OR, put it another way, is there another combination of development tools that can be shown to be significantly faster/higher performance.

And in both cases assume there would be thousands of concurrent users.
Posté le 04 juin 2018 - 21:57
Hi All,

Interesting discussion with lots of parameters and assumptions…

But I guess JP is talking about the WB application server hooked into the web server where the web server itself would be identical on the same hardware in the comparison. As WX is a framework/runtime based application server it is valid to say that other frameworks may be faster or slower simply because of the framework itself (optimisation for multi cores/clustering/JIT Compilation/paging data access/etc...).

In the open source world there are tons of web application frameworks available with tons of possible database backends and other middleware configurations. As you can see in charts like these here, where all these frameworks are benchmarked on the same underlying hardware, there are huge differences just because of the performance of the frameworks themselves.

As WB is so niche you probably won't find any independent benchmark of the WB/WX framework so I can only talk from experience and that is that it is (up to V21) remarkably slower than most other platforms I've used. But acceptable for most of my use cases though. We found bottlenecks specially when using HyperFileSQL CS with complex joins on the whole of teh WX platform. Moving the data to SQL server gave quite some performance benefits and boost. For WB, the page generation remained pretty slow in my opinion however. But that is just my subjective feeling of course because I never benchmarked it with hard figures.

But maybe this all has been largely improved with V23.


Just my <= V21 cents here
(Still having three V21 webapps running in WB),

Peter

Side note: In 2016 ASP .Net Core gave a 2300% performance gain over ASP .Net 4.6 for the Age of Ascent developers serving up to 1,15 million requests per second... Probably all written extremely well and efficient on a highly scalable platform but it indicates the importance of the runtime/application server (.Net Core vs .Net Classic in this case) to make the difference.
(Source: https://www.ageofascent.com/2016/02/18/asp-net-core-exeeds-1-15-million-requests-12-6-gbps/)