SpaceBase Can Now Be Used as a Spatial Redis | Hacker News Reader

SpaceBase Can Now Be Used as a Spatial Redis

blog.paralleluniverse.co

20 pointspron9 comments

I would kind of like for Redis to be used as a Spatial Redis. Would antirez consider adding an rtree/kd-tree/whatever-else type if a good pull request for one was available, or is that the sort of thing that's meant to be done "above" Redis (by, say, geohashing coordinates into scores for a zset)?

The comparison with Redis is totally misleading. SpaceBase is for completely different use cases and has a different underlying technology.

It is an interesting development though. It uses R-trees and the whole data set and index is keep in memory.

Product appears to have nothing to do with Redis and isn't even remotely compatible. Marketer uses buzzword to attempt to get credibility and hence immediately loses all credibility with actual market.

It can be used in the same way and for the same purpose Redis is used, only for spatial updates/queries. Take a look at the Thrift/Erlang documentation.

I don't think you understand Redis. Redis is also a key-value store, but implements a variety of structures, not just RTree, and has a large number of operations on those data structures (http://redis.io).

Spacebase appears to be a significantly more primitive KV store that implements RTree. Not that that's bad; looks interesting, even though it appears that the product won't be free, and so in my opinion is probably doomed.

Using 'Redis' in the title is just embarrassing for Spacebase and, by extension, for YC too. No need to troll the professional developer community with ridiculous comparisons.

I had the same impression. The author doesn't know Redis well or he wouldn't have made that comparison.

The comparison is meant to be functional - not technical. Most people use Redis as a low-latency key-value store either for write-heavy uses, short-term storage or as a cache. SpaceBase is, indeed, not a key-value store, but it can now serve the same purpose Redis does, namely, a low-latency, write-heavy data-store, only for spatial indexing rather than key indexing. I think I've made an apt, fair analogy.

Redis is a particular product, not a catch-all for low latency in-memory data stores. So saying your product is a Redis (which you did, rather than making an analogy) was merely embarrassing attention grabbing marketing palaver which does discredit to your product and suggests that you don't have the attention to detail and critical thinking skills required to properly implement a product in this space. See: MongoDB.

By your logic I could call memcached a "schemaless Redis" and it would be not only fair, but apt. Give me a break.

Reply on news.ycombinator.com