![]() Still need to find the right driver for cassandra, not going to happen today. I applied the same basic optimizations to both mysql and postgres. So I took some time to do this right today. I'm assuming they have better default management of the index, and it looks like they also have an interesting partitioning scheme. I wanted to see the baseline, and mysql's out of box speed was embarrassingly slow once the data size (or the index) got larger than available ram. I used the same schema for both but otherwise did not touch any setting or enable partitioning on either database (to make this work at all with the mysql database on our 'production' system I spent several days reading through various methods for improving performance and implementing them - limiting flush to disk, using >20 separate connections to insert data, partitioning, etc). ![]() The specific use case is for large time series data so I inserted data ~1000 rows at a time for 100s of GB of data. I kind of read the opposite, but fair enough. Also, from my reading, I wouldn’t expect MySQL to be slower than Postgres (except possibly for complex queries), so I think there must be something wrong with your MySQL benchmark.
0 Comments
Leave a Reply. |