It's my understanding that the amount of code is so small that they just have it all in ram (and maybe less accessed data goes to disk) and do a distributed grep.
Right, that's more or less what all the approaches I've seen use. It's not indexing in the sense of full-text indexing for fast retrieval, all of the data is more or less greped over. It's basically just a full-text search, not an index.
Comments
It's my understanding that the amount of code is so small that they just have it all in ram (and maybe less accessed data goes to disk) and do a distributed grep.
Right, that's more or less what all the approaches I've seen use. It's not indexing in the sense of full-text indexing for fast retrieval, all of the data is more or less greped over. It's basically just a full-text search, not an index.