How is this different than the many other columnar SQL databases and extensions?
Columnar querying is typical for OLAP, PostgreSQL engine is aimed at OLTP. This doesn't look like a good idea. Like adding side floats to a car and paddles to use it like a boat.
This goes against using the right tool for the right job.
How is 'jsonb' better than mongodb ?
Because you can use the same tool, you'll support the same db, you'll pay 0$ for licensing, you may pay 0$ for sharding (postgresql-xc) etc.
Also, I bet VoltDB, a modern open source OLTP, can beat this thing hands down. Also in-memory and clusters. Complex store procedures precompiled and many other goodies.
Commercial column stores like Vertica should be orders of magnitude faster.
Don't know why you are being down voted, but you should understand its about tradeoffs.
If I want to increase my database performance I can either
1.) Build this plugin and integrate it into my already working ecosystem
2.) Spend time researching, testing, and deploying VoltDB.
Given the popularity of postgres, and the relatively low friction solution of (1.), its clear why this could be an adequate solution. Sure you won't be as fast as VoltDB, but as an outside engineer we don't know the potential customers requirements, and if being as fast as VoltDB actually matters.
Comments
How is this different than the many other columnar SQL databases and extensions?
Columnar querying is typical for OLAP, PostgreSQL engine is aimed at OLTP. This doesn't look like a good idea. Like adding side floats to a car and paddles to use it like a boat.
This goes against using the right tool for the right job.
How is 'jsonb' better than mongodb ? Because you can use the same tool, you'll support the same db, you'll pay 0$ for licensing, you may pay 0$ for sharding (postgresql-xc) etc.
Downvotes? Lovely.
Also, I bet VoltDB, a modern open source OLTP, can beat this thing hands down. Also in-memory and clusters. Complex store procedures precompiled and many other goodies.
Commercial column stores like Vertica should be orders of magnitude faster.
Don't know why you are being down voted, but you should understand its about tradeoffs.
If I want to increase my database performance I can either
1.) Build this plugin and integrate it into my already working ecosystem 2.) Spend time researching, testing, and deploying VoltDB.
Given the popularity of postgres, and the relatively low friction solution of (1.), its clear why this could be an adequate solution. Sure you won't be as fast as VoltDB, but as an outside engineer we don't know the potential customers requirements, and if being as fast as VoltDB actually matters.