Magento Performance: Enterprise vs Community
Thursday, August 26th, 2010 by
There’s an opinion that Magento platform is something slow and performs not really well. That’s why a lot of enterprise customers have serious doubt about establishing their online business based on Magento platform. We keep working hard on performance optimization solutions for Community Edition and we’ll post another article with the results we were able to achieve at the moment, but today let me share some numbers related to Enterprise Edition performance.
Everybody knows that one of the most important benefits of Magento Enterprise Edition is that it’s really fast. Much faster than Community version, and we saw how fast it is in the past while we were working with EE . The thing is that we have never tested the platform to determine how exactly fast the EE is. Well, guys, it is unbelievably fast.
How did we test
First of all, I need to stress that we took a home PC to make the test environment. That means that your site will perform even better on professional hosting. But of course you’ll need a dedicated machine for your hosting solution. Again: we are talking about enterprise customers, remember? You need to realize – there’s no way to make millions spending $20 to hosting.
Here’s the hardware specification:
CPU: 3.16 GHz Core 2 Duo E6500 4 Gb DDR2 RAM SATA 2 HDD drive OS: 64X Debian 5 Lenny Linux
We ran tests using Siege 2.69 with various number of concurrent users, each making 10 hops based on prepared path list. We judge test as failed once more than 5% of transactions lost.
Default Apache installation
We just installed LAMP and deployed Mangeto versions there. There was no tuning at all.
Nothing unexpected with Community Edition: 100 users test is failed. But take a look how good EE is:
Enterprise edition with 100 users performs better than community edition with 10.
Now, let’s tune the server a little. We have modified Apache preferences and set up PHP APC.
Tuned Apache + APC
And now you can see the real difference. Finally CE was able to handle a hundred. At the same time the gap became really huge. EE has no problems with 200 and moreover 300 users. Good numbers.
Three hundred users: EE still does it for less than a second
Isn’t great? I mean 300 concurrent users is a good target for even a big store. The exact response time for 300 spartans is .72 sec.
So what we have as a bottom line? Performance is not a problem for Enterprise users. It performs pretty good even with a little environment tuning. We actually keep working with optimization for EE, I feel we can definitely achieve less than a half-a-second and more.
And yes, we also have some good numbers with CE, but it’s another story.
Special thanks to Iwan Tarasenko @ Oggetto Labs.

I would say the biggest difference I’ve noticed between community and enterprise: COMMUNITY EDITION ACTUALLY WORKS!
Seriously, I can send you all of the tickets I’ve put in to enterprise support about bugs and broken functionality that otherwise work in the community edition. I can also send you all of the patches they responded with for me to go and apply. It’s ridiculous and scary to be paying $12K/YEAR for software that doesn’t work as well as it’s free counterpart.
So frustrating having to bring my development to a screeching halt due to their laziness and shortcomings. Makes me look bad to my overseers as well when I can’t meet deadlines.
That is not true.
First of all Enterprise actually works. We’ve built several websites on this awesome platform and everything was just cool. Every single feature works good. I’m talking about at least 1.8 and 1.9 versions. And for enterprise clients who have significant traffic and sales $12K is a fair price. I believe so.
We noticed the same thing about community vs enterprise.
However, since we were in the same community boat that most people are, we decided to develop the our own whole-page caching solution so that we could achieve the same concurrency and page-load times as EE enjoys. That product is called Lightspeed (Not to be confused with the web server).
Check it out here – http://www.tinybrick.com/magento-modules/performance/improve-magentos-slow-performance.html/
Here’s some benchmarking we did on a recent site with our Lightspeed module installed – http://www.tinybrick.com/var/product/doc/Lightspeed_Benchmarks.pdf
Nice to know. We’d be really interested to try this. Is there anyway we can get a version for non-commercial use?
Chris, if your lighty product is so good, why is your site running nginx?
Very good share! Good work on comparing the performances of two editions.
so what is making EE so fast?
if i google whats the difference between those 2 editions, i found just that they are using same core, but EE have some enterprise modules.
does that mean that EE have some special caching module?
there are couple of caching modules out there what can be bought even for CE, do you have any experience with it (can you recommend some what will be closest to EE version in terms of performance)?
ta
Well, most of all it’s FPC – full page caching mechanism.
Yes, for Community Edition there are several solutions that apply such thing as well.
I can recommend Tiny Brick’s Light Speed module, and Nitrogento by NBS.
You can also try to integrate 3rd party solution (on the higher architect level) – such Varnish, google Varnish + Magento.
The thing is that it takes addition development effort to integrate those modules because Community Edition itself is not compatible with full page caching and and by default it will cache all pages, even though you don’t need particular blocks caching. So you’ll need to put extra effort on punching the “holes”. Especially if you use community modules, it’ll rise a hell on your website.
I know this is a bit late, but:
What were the versions used in these tests?
And could You specify what Your “LAMP” in this case was, since different distributions have different default configurations for Apache, PHP and MySQL.
Also, what kind of tweaks were made?
Hey Toni.
To be honest, the post is pretty much outdated. Today we use a bit different technologies for test and optimization.
We’ll probably update the entire article early next year.
Merry Christmas and Happy New Year.