Indeed LSM Tree is a fast key/value store that is efficient for high-volume random access reads and writes.
Indeed uses it for many applications including serving job data for 100s of millions of job searches each day.
Indeed LSM Tree is composed of three parts:
A typical use case is requiring a key/value store replicated on multiple servers. Instead of replicating the store itself, we store writes in record logs which are slaved to each server. Each server then builds its own LSM tree from the record logs.
10,000,000 operations using 8 byte keys and 96 byte values
Random writes:
Software | Time |
---|---|
Indeed LSM Tree | 272 seconds |
Google LevelDB | 375 seconds |
Kyoto Cabinet B-Tree | 375 seconds |
Random Reads:
Software | Time |
---|---|
Indeed LSM Tree | 46 seconds |
Google LevelDB | 80 seconds |
Kyoto Cabinet B-Tree | 183 seconds |
Same benchmark using cgroups to limit page cache to 512 MB
Random Reads:
Software | Time |
---|---|
Indeed LSM Tree | 454 seconds |
Google LevelDB | 464 seconds |
Kyoto Cabinet B-Tree | 50 hours |
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。