Social network and blogging platform LiveJournal has been around a long before Facebook came to be. In the early years of its existence the company had to implement an invite code system that curbed its user base, simply because its infrastructure couldn’t keep up with membership growth.

A lot has changed since then. LiveJournal expanded to hundreds of servers, but as the environment grew more complex similar issues associated with scalability and data protection started to emerge. The company developed its own software to tackle these issues, and eventually open-sourced it – a move that helped kick start a whole new trend in the enterprise. All these factors make the firm a very good example for an entrepreneur who’s looking to start a web-scale business.

LiveJournal had to solve a lot of challenges before it could get its operation going at full swing. Achievement persistent connection was an issue, along with single points of failure, and other performance related barriers.

To solve these the social network took a parallel approach, and removed bottlenecks by spreading out reads and writes, splitting up jobs, allocating storage more efficiently using DRBD and improving caching. It achieved the latter by creating a tailor-made solution called memecached, an in-memory key-value store that has picked up a lot of traction in the open-source world.

The hard work of LiveJournal’s internal IT department (including Brad Fitzpatrick, who discussed this topic more extensively in a session founded here) produced several tips worth taking into consideration; parallelization, for one, is key to scaling. And so is taking care of the low level issues that have a tendency of rearing their ugly heads once a cluster has reached its critical mass.

Lastly, the engineers also learned firsthand that getting a better feel for the purpose a system is supposed to serve – powering a social network in this particular case – makes things easier across the board.