Massive matching speed increases
June 24th, 2008
The past week we have been re-architecting our matching system to make it dramatically faster.
When we started Toluu the matching system ran on a small set of data and, like most code running against small sets of data, it ran very fast. However, as Toluu has grown we have had to make changes. A few weeks back we optimized the way we accessed the data in the database, this sped up the process by 10x or so. But we knew this was only a patch.
We then began on the work we just launched. We have moved to a much more scalable system that will last for many doublings in size of the userbase.
We are so excited about these optimizations that we will be spending the rest of this week making optimizations across the rest of the site. These optimizations will lay the groundwork for a new set of features and our growing userbase.
