Its great to see Piwik so active. I've been using it since version 0.xsomething. Its realiable, works great with my websites and I really like having an alternative to Analytics with a nice UI that I can host myself.
Cheers Piwik, keep doing what you are doing. And all the best with the new marketplace.
I love piwik, I find it way easier to use than google analytics. One thing interesting in it, nowadays: you have direct access to data in database. This is priceless to find data to feed neural networks.
One thing I liked about Piwik was their API documentation. Short and simple, and the examples are using actual data, meaning you will get a url of that API returning the live data in your database for you to inspect, right from the documentation page. Of course all of this is possible because you are self-hosting the Piwik on your server, but it is a good feature that I really liked.
This is actually the first time I've heard of Piwik but it seems like a great alternative to Google Analytics/Mixpanel/etc.
Kudos to them for also finding a business model alongside the open source application. That set up usually lends itself to long running OSS as there is (hopefully) always money coming in.
Piwik does scale to millions of pages per month. We've managed installs of Piwik at _1 billion_ pages per month that's 30 million per day! Don't believe what you read on Hacker News ;-)
How it handles that traffic? It would be great if they could share the architecture because using Mysql for analytics is not easy, you can't expect Mysql to handle concurrent queries that needs to aggregate billions of records when I visit the dashboard.
No, not usable in real life. Old-school PHP bloat that will slow down your site. Bonus: breaks occasionally on updates. Developers not willing to learn techniques needed for a quick logger with minimal impact. Run away from people using this with high traffic sites in a still-2005-PHP-mindset.
Piwik is a great alternative to Google analytics, can't recommend them enough. On my websites I really like the option to track visits directly from the log file. It isn't as detailed as js tracking, but gives accurate number of visits (not altered by ghostery / privacy plugins).
I'm using Piwik for JS tracking (pageviews, searches and external links) and Goaccess for apache log (http statistics updated every 3 hours), great combo!
Heads up that the goaccess in Ubuntu's universe repo is woefully out of date, so DON'T apt-get install it. Download and build from source - you'll get a much better experience.
I've noticed this with several other packages in universe recently too, it seems to be updated much less than you'd expect. It's a community-maintained repo, but you'd think the Ubuntu community is large enough to stay on top of some of this stuff. As it stands, the redis in universe on 14.04 has had unpatched security issues for at least 2 years.
Piwik is an awesome tool. But today I'm only using their JS Tracker to send requests to a pool of nginx servers.
The event logs are parsed with Fluentd and saved in a mongodb instance. After that, a proccess do some aggregations and put in my postgresql (app db).
We have this setup because we run our custom and specific analytics for a lot of clients.
Do anyone have a similar setup and have some experiences or hints to share?
45 comments
[ 2.9 ms ] story [ 72.6 ms ] threadCached version: http://webcache.googleusercontent.com/search?q=cache:http://...
For those that don't know: Piwik is a open source Google Analytics alternative.
Cheers Piwik, keep doing what you are doing. And all the best with the new marketplace.
Kudos to them for also finding a business model alongside the open source application. That set up usually lends itself to long running OSS as there is (hopefully) always money coming in.
My Piwik instance is nearly unusable because any non-trivial request takes ages.
For scaling tracking, QueuedTracking plugin is the solution: http://piwik.org/docs/setup-auto-archiving/
For managing DB size, see: https://piwik.org/docs/managing-your-databases-size/
For any other issue, maybe report in Github tracker: https://github.com/piwik/piwik/issues
If that isn't enough http://piwik.org/docs/optimize-how-to/ shows you how to configure php and mysql for more performance.
Actual hardware requirements are quite small for small sites (say < 10k visitors/day). Any 10-15$ VPS should be fine.
We stopped using MySQL.
Or to word it differently, we stopped using Piwik. They lived together, they died together.
(Quick test: if you're counting page views in millions per day, you're gonna have a tough time).
If you want Log File Analytic, you should definitely give GoAccess a try.
https://goaccess.io/
I've noticed this with several other packages in universe recently too, it seems to be updated much less than you'd expect. It's a community-maintained repo, but you'd think the Ubuntu community is large enough to stay on top of some of this stuff. As it stands, the redis in universe on 14.04 has had unpatched security issues for at least 2 years.
Piwik certainly has more features though.
It wouldn't be all that hard to set up an Elasticsearch/Kibana instance and then throw your maillogs into it.
We have this setup because we run our custom and specific analytics for a lot of clients.
Do anyone have a similar setup and have some experiences or hints to share?