Searchy

Searchy is a fast and flexible indexing and search tool. Actually, being "fast and flexible" is the goal of Searchy (and the rest of the software from my research group). You have to decide for yourself if Searchy has achieved its goal or not. Searchy is written mostly in C++ and has been tested on FreeBSD and Linux.

Download

searchy-0.1.tar.gz It requires Berkeley DB 4.0+.
To use it and extend it, check out Searchy FAQ

Report bugs?

Who's using searchy?

Thanks to ...

Benjie Chen
Jeremy Stribling

Other search engines out there

I have done some simple benchmark performance tests with other search engines.
The test corpus is my own Mail archive, 7534 files, 385 Mbytes.
The test machine is a FreeBSD box with 3.06GHz CPU and 1G memory

Index time Search Time (w/o context display)

Searchy 120 s 5-10ms

* Searchy indexes all the words, generating a 397M index file with 6 bytes per word.

Lucene v1.4 211 s 6-100 ms

* Very strangely, Lucene has missed a lot of words. So its index file is not complete (43M) and the real index time could be longer.

Glimpse v4.18 1.5 hour+
* Glimpse indexes all words, generating a 132M index. Its indexer uses less than 10M memory, so I suspect one can tune to use more memory for faster indexing

[Home] [Research]

	Index time	Search Time (w/o context display)
Searchy	120 s	5-10ms
* Searchy indexes all the words, generating a 397M index file with 6 bytes per word.
Lucene v1.4	211 s	6-100 ms
* Very strangely, Lucene has missed a lot of words. So its index file is not complete (43M) and the real index time could be longer.
Glimpse v4.18	1.5 hour+
* Glimpse indexes all words, generating a 132M index. Its indexer uses less than 10M memory, so I suspect one can tune to use more memory for faster indexing