What's the best way to keep a large list (e.g. 10K items) in Redis, where I also want to efficiently retrieve items by key.
It seems Redis has no data structure equivalent to Java's OrderedHashMap, which accomplishes this, so maybe it's necessary to maintain a set and a list and ensure they stay in sync.
Use a sorted set;
Add some bookmarks; use current time for score to sort chronologically:
> zadd bookmarks 123 "bk1"
> zadd bookmarks 456 "bk2"
> zadd bookmarks 789 "bk3"
> zadd bookmarks 999 "bk4"
To get a bookmark, you need the index first:
> zrank bookmarks "bk3"
> "3"
...then pull the bookmark by index:
> zrevrange bookmarks 3 3
> "bk3"
If you don't want to use timestamps, you can sort bookmarks lexicographically using "1" for score:
> zadd bookmarks 1 "link_xyz"
> zadd bookmarks 1 "link_abc"
> zadd bookmarks 1 "link_foo"
> zrange bookmarks 0 -1
1) "link_abc"
2) "link_foo"
3) "link_xyz"
The index lookup is O(log(n)), add to that O(log(n)+1) to pull a single member by index; better than O(n) for lists.
Also, if you add the same bookmark twice, redis automatically replaces the previous member, so you avoid duplicates.
Hope it helps,