People often want to know how Xapian will scale. The short answer is "very well" - a previous version of the software powered BrightStation's Webtop search engine, which offered a search over around 500 million web pages (around 1.5 terabytes of database files). Searches took less than a second.
The largest recent installation we're aware of is probably gmane, which currently indexes over 50 million mail messages.
One effect to be aware of when designing benchmarks is that queries will be a lot slower when nothing is cached. So the first few queries on a database which hasn't been searched recently will be unrepresentatively slow compared to the typical case.
In real use, pretty much all the non-leaf blocks from the B-trees being used for the search will be cached pretty quickly, as well as many commonly used leaf blocks.
In a large search application, I/O will end up being the limiting factor. So you want a RAID setup optimised for fast reading, lots of RAM in the box so the OS can cache lots of disk blocks (the access patterns typically mean that you only need to cache a few percent of the database to eliminate most disk cache misses).
It also means that reducing the database size is usually a win. The Flint and Quartz backends both compress the information in the tables in ways which work well given the nature of the data but aren't too expensive to unpack (e.g. lists of sorted docids are stored as differences with smaller values encoded in fewer bytes). There is further potential for improving the encodings used.
Another way to reduce disk I/O is to run databases through xapian-compact. The Btree manager usually leaves some spare space in each block so that updates are more efficient (though there are heuristics which will fill blocks fuller if they detect a long sequence of sequential insertions so adding documents to the end of an empty database will produce fairly compact tables, apart from the postlist table). Compacting makes all blocks as full as possible, and so reduces the size of the database. It also produces a database with revision 1 which is inherently faster to search. The penalty is that updates will be slow for a while, as they'll result in a lot of block splitting when all blocks are full.
Splitting the data over several databases is generally a good strategy. Once each has finished being updated, compact it to make it small and faster to search.
A multiple-database scheme works particularly well if you want a rolling web index where the contents of the oldest database can be rechecked and live links put back into a new database which, once built, replaces the oldest database. It's also good for a news-type application where older documents should expire from the index.
You could take this idea further by implementing an ultra-compact read-only backend which would take a flint Btree and convert it to something like a cdb hashed file. The same information would be stored, but with less overhead than the flint Btrees (which need to allow for update). If you need serious performance, implementing such a backend is worth considering.
The flint backend (which is now the recommended backend) stores the indexes in several files containing Btree tables. If you're indexing with positional information (for phrase searching) the term positions table is usually the largest.
The current limits are:
So what is the limit for a modern OS? Taking Linux 2.4 with the ext2 filing system as an example, quoting from linux/Documentation/filesystems/ext2.txt:
Filesystem block size: 1kB 2kB 4kB 8kB File size limit: 16GB 256GB 2048GB 2048GB Filesystem size limit: 2047GB 8192GB 16384GB 32768GB There is a 2.4 kernel limit of 2048GB for a single block device, so no filesystem larger than that can be created at this time. There is also an upper limit on the block size imposed by the page size of the kernel, so 8kB blocks are only allowed on Alpha systems (and other architectures which support larger pages).