How big is your Elgg network?


I'm working for a team that runs a fairly large Elgg network.  We're currently working on scaling it up to accomodate more users, and as part of our research, I'd like to pose a question to the community:

How active is your Elgg community?  How many users, actions, etc?

Our Elgg community currently has over 35,000 users who participate in over 1,800 different groups, and we're curious to know how large our fellow Elgg users' communities have become and what kind of activity and performance you've had.

We average something like 200 simultaneous users during business hours, and our network often performs up to 300 actions/second.

I'd love to hear about other people's Elgg experiences and how it performs for your user base!

  • I am running an Elgg community with 22.000 users on a single machine. It runs fine, but have literally done and tested everything that has been posted on this community and elsewhere on tuning Elgg Code and LAMP. A lot of it makes sense, but some have hardly any benefit (like running Nginx instead of Apache or use memcached).

    An interesting topic I am currently investigating is to see if I can get Elgg runing on Facebooks Hip Hop Virtual machine (HHVM). Until now it has been a drag getting it compiled and running and so far the fastcgi interface is not doing what it is supposed to do. But the claims are so fantastic on the performance benefit that I am struggling on until I find it out if that is true for Elgg also.

    Will let everyone know how that went in this group. But even if it does work, it is not something you should have to consider for 35.000 users if you need to hire engineers on getting it done. Then just buy another machine and split of the db.

    Well, if you have any nice tips on what you have done. Let us know too

  • FYI Currently Elgg won't run correctly on HHVM, where are at least some core changes required, but I didn't manage to run it in stable manner. See for the details and discuission so we don't hijack this thread.

  • I don't have much active users but I am getting non registered members viewing blogs. As I'm typing this, Google Analytics is reporting that there are 486 active visitors at once. It almost peaked at 1k anonymous users the other day and the CPU has been fluctuating around 10-25%. I guess that's not so bad. 

    Also Google reports that my site is getting over 100 unique hits a minute every single day so my server is getting worked!

  • I manage a number of sites ranging from a few dozen to over 100k registered users.  Elgg can certainly be scalable, just not necessarily out of the box, though work on that front is always improving.

    If you search around there's already been lots of suggestions made for how to do this, I think someone may have even aggregated them into some sort of informal document somewhere (if not they should).

    But things like changing the db tables to Innodb, having the database on a separate dedicated machine with high IO, and using a dedicated search index (plug: are a big help.  Elgg's default search is the hardest hit to the database in my experience, offloading that is a *huge* help.

    Unfortunately I don't have any hard metrics for you though, so consider this just good old anecdotal evidence.

  • I'm currently running a custom elgg application. It has 100 students registered and 84 active (some dropped out).

    So far there are 5 classrooms (groups). Yesterday I had 39 students connected at the same time, working on an assigment, on a shared server, with no performance issues.

    Plugins we use

    Files Plugin


    Roles Framework

    Roles for Profile Manager

    Student Role

    Teacher Role

    SW Social Docs

    Custom Twitter Bootstrap Theme

    Groups Plugin

    TinyMce Pro


  • Thanks so much, folks!

    Matt Beckett: We've noticed the search is rough on the servers - we're intending to offload to a Google Search Appliance pretty soon.  Our other major issue is with mass messaging.  Some of our user groups have up to several thousand members and things like emails for group notifications are putting up a huge load - although we currently have 4 servers running our Elgg network, so we're coping.  

    Our performance is presently pretty good - just a touch of slowdown during peak hours.  Soon we'll be running testing with HP Loadrunner software to test the performance of high-priority transactions with 5,000 virtual users - I'd be happy to share the results as we'll be learning which transactions are the most troublesome and potentially some fixes for them.

    Thanks for the responses! 

  • The new notification system in 1.9 resolves this mass group notification issue.

  • Look forward to hearing the results of your tests.

    The place to aggregate performance recommendations is here:

Performance and Scalability

Performance and Scalability

If you've got a need for speed, this group is for you.