Why is revo so slow?

Hi!

As RC-2 is out, I tried it again, but it still feels much slower than Evolution. I use a shared web server and you can really feel the difference. Some data on nearly the same site (yams is active on Evo)

Revolution:
MySQL queries 0.0002 s for 1 queries(s), document parsing 1.0864 s, for a total time of 1.0866 s, and retrieved from cache.

Evolution:
MySQL queries 0.2556 s for 44 queries(s), document parsing 0.6932 s, for a total time of 0.9488 s, and retrieved from database.
or
MySQL queries 0.0010 s for 3 queries(s), document parsing 0.1760 s, for a total time of 0.1770 s, and retrieved from cache.

As you can see Evo is faster on retrieving directly from database as revo is on retrieving from cache. Evolution feels like getting served static html files where Revolution thinks about it that extra second.

Do that test with 1000+ Documents, 30+ snippets, and other customizations on each platform. Then come back.

Revo outperforms Evo signficantly on larger sites, especially as Evo’s cache file increases. Revo also has quite a bit more power under the hood.

Think of it like a graph: Evo’s performance starts very low but slopes upward as the site grows, Revo’s starts slightly higher than Evo, but performance stays mostly flatlined as it grows.

Also, try putting memcache with Revo, and then, well, yes.

Quote from: splittingred at May 31, 2010, 09:55 PM

Think of it like a graph: Evo’s performance starts very low but slopes upward as the site grows, Revo’s starts slightly higher than Evo, but performance stays mostly flatlined as it grows.

Is there some data where the break even is? At wich site size Revo starts outperforming Evo?

Also, try putting memcache with Revo, and then, well, yes.

Thats the problem with shared hosting...

Quote from: fthieme at Jun 01, 2010, 12:10 AM

Quote from: splittingred at May 31, 2010, 09:55 PM

Think of it like a graph: Evo’s performance starts very low but slopes upward as the site grows, Revo’s starts slightly higher than Evo, but performance stays mostly flatlined as it grows.

Is there some data where the break even is? At wich site size Revo starts outperforming Evo?

No, but feel free to run some benchmarks.

Quote from: fthieme at Jun 01, 2010, 12:10 AM

Also, try putting memcache with Revo, and then, well, yes.

Thats the problem with shared hosting...

Scalability starts at the hosting level. If you are using shared hosting, you are not worried about scalability in the first place. In many cases, shared hosting users should probably stick with Evo if it makes more sense. The size and complexity of the site will likely be the determining factors there.

Quote from: OpenGeek at Jun 01, 2010, 08:14 AM

Scalability starts at the hosting level. If you are using shared hosting, you are not worried about scalability in the first place. In many cases, shared hosting users should probably stick with Evo if it makes more sense. The size and complexity of the site will likely be the determining factors there.

So Revo is aiming on the large scale users as Evo still fits the need for smaller sites. I have no problem with that as long as Evo won’t be droped in favour of Revo asap..

Quote from: fthieme at Jun 01, 2010, 08:56 AM

So Revo is aiming on the large scale users as Evo still fits the need for smaller sites. I have no problem with that as long as Evo won’t be droped in favour of Revo asap..

Sort of, but I was more trying to illustrate a point that the architecture of Revo is more suited for scalability than Evo ever can be. Development and support for Evo will continue for a long time to come, so don’t worry about that, but most new features and innovation will be taking place in Revo going forward. For non-complicated, baseline sites, Evo may be slightly faster (disclaimer: this depends completely on the Snippets and other implementation specific details of your site) and perhaps slightly less resource intensive in some situations. But the capabilities are much more limited, and if you ever need to think about scaling up, doesn’t it makes sense to implement on the platform that is capable of doing that in the first place?

I really think some different benchmarks on a variety of sites in a variety of environments could help better identify the actual metrics for making such a decision when shared hosting limitations are an issue, but it’s definitely not a line drawn clearly in the sand. There are simply too many variables to consider. But I will say that if you are stuck with a shared hosting platform that is not kept up-to-date (i.e. isn’t running most recent 5.2.x or 5.3.x versions of PHP; Revolution runs more efficiently with each new release), Evolution may indeed be a better choice for your project at this time.

I’ll have to chime in here as well, at least regarding the Manager interface. In Evo, when I open a doc for editing it’s immediate. In Revo, it’s 6-10 seconds. I think this is ExtJS? Also, it appears to repaint the entire screen which is annoying, especially the tree which "re-accordians". I have two clients with 100-200 documents in Evo and yes, the Manager tends to be a little slow. I would add that the "cleaning up" in Evo is also annoying but not as bad as the Revo performance.

Is it ExtJS that’s causing the Manager to behave this way?

What version of Revo are you running? A lot has been done on this front.

Quote from: maclonghorn at Jul 01, 2010, 07:58 AM

I’ll have to chime in here as well, at least regarding the Manager interface. In Evo, when I open a doc for editing it’s immediate. In Revo, it’s 6-10 seconds. I think this is ExtJS?

Ext JS is like a truck: powerful but slow! It will be faster while using more size of RAM.

I find the browser helps, on the same machine revo is faster in Chrome than FF, because of the better js parsing in Chrome. Sometimes revo is noticeably slower in FF, its not ’slow’ overall certainly not like the times you are seeing, just slower. As I’ve posted before on my main dev box using revo feels the same as using a native C++ app.