[HN Gopher] CockroachDB vs. Scylla Benchmark
___________________________________________________________________
CockroachDB vs. Scylla Benchmark
Author : northstar702
Score : 23 points
Date : 2021-01-21 18:47 UTC (4 hours ago)
(HTM) web link (www.scylladb.com)
(TXT) w3m dump (www.scylladb.com)
| PeterCorless wrote:
| Beyond raw benchmarking, here's an interesting use case from
| StadiaMaps. https://www.scylladb.com/2020/07/30/stadia-maps-
| using-scylla...
| atombender wrote:
| I wonder why that app needed a relational database for vector
| tiles in the first place. That seems like read-only data that
| could be heavily cached at the edge, too. I'm not surprised
| that Scylla outperforms CockroachDB there, though.
| SR2Z wrote:
| It probably has something to do with the difficulty of
| maintaining a worldwide spatial index otherwise - it's
| really, really hard to keep the index consistent and up-to-
| date with a bespoke product, and every time I've worked on
| GIS the solution for more than a few square km of data has
| been "stick it in a database so we can index it."
| atombender wrote:
| But tiles don't need to be spatially indexed -- they're
| offsets on a 2D plane in a Mercator projection. I've only
| worked with bitmap tiles, but it's my understanding that
| vector tiles use the same system. So all you need is fast
| (x, y) lookup.
___________________________________________________________________
(page generated 2021-01-21 23:02 UTC)