1

Why are LSM trees called index structure? My understanding is that they are a "storage technique" i.e., how you store the data underneath.

In fact, for performance reasons you would need an index on top of this data --in this case of LSM an in-memory hash map or a block index - to quickly locate the data you want.

I understand that you can store the index content as LSM tree internally but almost everywhere I've read LSM tree being referred as indexing structure. I find this confusing and doubt if I've got the concept right?

Thanks

biqarboy
  • 852
  • 6
  • 15
user2599672
  • 11
  • 4
  • Perhaps you can state what *you* think defines an index structure, if you don't think LSM qualifies. – Damien_The_Unbeliever Aug 05 '18 at 08:11
  • Hi @Damien_The_Unbeliever. Thanks for the response. Per my understanding, index is an additional structure that's created to make it easy to get to the actual data that is indexed. As I mentioned above, an index can be implemented as LSM tree, I didn't quite get why it's referred to as index structure. It's just a way to store the data which supports high write throughput as it aligns well with the hardware (disk friendly) – user2599672 Aug 05 '18 at 08:52
  • 1
    "a way to store data" sounds pretty well like a definition for the word "structure". So, we agree that an LSM can be used for indexing, and we're describing a structure. What would you call it? – Damien_The_Unbeliever Aug 05 '18 at 08:54

0 Answers0