How Confluence is used in SAP

I want to talk a little bit about how Confluence is used in SAP: www.sdn.sap.com/irj/scn/wiki

For those who do not know, a few words about what it is. Konfa, as it is often called, is a traditional wiki system . For each article keeps a list of changes in a convenient form: you can see what changes Masha made a week ago, and what Peter changed six months ago. The chip is in the convenience of comparing different versions of the article. Developed workflow systems also store different versions, say the doc file, but they are inconvenient to compare. It is necessary, for example, to highlight your comments in a different color. This is often not required on the wiki. On the example of Wikipedia, it is clear that this is the most convenient way to store and collectively edit a certain system of knowledge (articles).

Confluence paid . Prices vary widely from 10 bucks for 10 users, to $ 12,000 for wirewashing companies. Free to open source projects, charitable organizations ( Non-profits and Open Source projects ).
')
You buy a license - you get the source. You can see how everything is arranged inside, change something, if necessary.

Written in Java. Written well, uses good technology, but not always fresh versions. In our (ver. 3.2.1) is used:

• Java 1.5;
• Spring 2;
• WebWorks;
• Hibernate 2;
• search - Lucene 2;
• the interface draws the Velocity Template Language (sometimes Freemarker is also used, which causes some confusion);
• Javascript: JQuery 1.3.2 (plus a bit of DWR).

Confluence 4.0 was recently released, it is possible that more modern technologies are used there, I have not had time to look at it myself. I hope the comments prompt. I know that 4-ka already uses HTML5.

The system is developed by the Australian company Atlassian. Judging by the photos from the office , by their good products, this company should be pleased to work.

Confluence is widely used , including Adobe Systems, Bloomberg, Cisco Systems, IBM, and, of course, SAP AG.

How are we used

Architecture

I will try without details, so as not to violate the NDA.
Physically, there is 1 powerful multi-core server with 16 GB of RAM. It contains a cluster of 4 virtual machines. On each Linux machine, the NetWeaver application server 7.2. The cluster is easily assembled with built-in Confluence.
The database has its own SAP-ovsky, MaxDB, under it 2 servers are additionally allocated: active / passive.
Current attendance can not say, I know that only about 2 million registered users.
Akamai caching CDN is used.

Improvements

Made single sign-on (SSO), the list of users and groups is taken from the corporate LDAP.
Also made a number of improvements in the original system. The main directions of improvements:

• loading of user files, various restrictions added, who can download which files, etc.
• virus check
• working with groups, synchronizing them from LDAP, caching,
• validation of articles and comments
• binding to the scoring system,
• sending additional e-mail notifications.

A number of plug-ins have been written for convenience: various decorations, formatting, article sorting, and more.

Features

Works cheerfully, copes with a large number of users.
Stable. I can compare with the Jive Forums, my colleague deals with them: there are always problems, improvements, a headache, although it is difficult to compare different products.

Alternatives

The first thing that comes to mind is the most famous wiki engine - the Media Wiki . Free, open source, written in PHP.
Cool comparison Confluence Vs Mediawiki .
Still known alternatives: Redmine , Trac . This is an open source system, written in Ruby and Python, respectively. I did not use them myself, so I can not comment.

Personal impressions

Pleasant system, adequate support from Atlassian. Perhaps a too complex plugin system, a beginner is not easy to understand. Well, fresher libraries are desirable to use.