3 ms·

I love Postgres, but lack of parallel query processing (same query leveraging multiple threads) has been a deal-breaker for some projects. Thankfully, Postgres

by tresil 10y ago

I love Postgres, but lack of parallel query processing (same query leveraging multiple threads) has been a deal-breaker for some projects. Thankfully, Postgres is finally starting to make some progress here. Also of note, In November MS added a number of previously enterprise-only features to all versions of SQL Server 2016 (Service Pack 1 and later). Examples: In-memory OLTP, ColumnStore indexes, are available even in the free Express edition now. Native Postgres, to my knowledge, doesn't have comparables for these yet. https://blogs.msdn.microsoft.com/sqlreleaseservices/sql-server-2016-service-pack-1-sp1-released/ https://blogs.msdn.microsoft.com/sqlreleaseservices/sql-serv...

tracker1 10y ago
I've always found MS-SQL pretty impressive in terms of features/value, though they get pricey pretty quickly, I've always found it nicer to work with than Oracle or DB2 by comparison... Although, I do wish they'd support the more common LIMIT syntax that many other DBs use. That said, from a devops perspective MSSQL doesn't have any real peers. Some of the analytical tooling around MSSQL is pretty good too. Have had to track performance issues in the past. Even the general query analyzer integrated with Management Studio is great. I do hope they consider going the electron route for cross-platform management studio, it's worked really well for VS Code, and can imagine it would be a great fit for this as well. Even if quite a bit outside their current tooling.
dspillett 10y ago
> I love Postgres, but lack of parallel query processing ... for some projects. It has surprisingly little effect on many workloads though - then engine is fairly conservative on when it will use it too (and there are easy ways for you to accidentally make it not consider parallelism). For many workloads you are often better off using the many CPU cores to run different queries than cooperate on a smaller number. Data warehousing and "big data" are a place where this isn't the case though, and other features MS SQL Server has over PG and others shine here too: the column store indexes, in-memory processing, and so forth. And you get these in the free "express" editions too, as of 2016sp1, though that is only really useful for small projects and prototypes due to other limitations in that edition (10Gb/database, max 4 CPU cores & 1.5Gb RAM put to use per instance) so you'll be paying for standard edition at least for real work.

by tresil 10y ago

tracker1 10y ago

I've always found MS-SQL pretty impressive in terms of features/value, though they get pricey pretty quickly, I've always found it nicer to work with than Oracle or DB2 by comparison... Although, I do wish they'd support the more common LIMIT syntax that many other DBs use. That said, from a devops perspective MSSQL doesn't have any real peers. Some of the analytical tooling around MSSQL is pretty good too. Have had to track performance issues in the past. Even the general query analyzer integrated with Management Studio is great. I do hope they consider going the electron route for cross-platform management studio, it's worked really well for VS Code, and can imagine it would be a great fit for this as well. Even if quite a bit outside their current tooling.

dspillett 10y ago

> I love Postgres, but lack of parallel query processing ... for some projects. It has surprisingly little effect on many workloads though - then engine is fairly conservative on when it will use it too (and there are easy ways for you to accidentally make it not consider parallelism). For many workloads you are often better off using the many CPU cores to run different queries than cooperate on a smaller number. Data warehousing and "big data" are a place where this isn't the case though, and other features MS SQL Server has over PG and others shine here too: the column store indexes, in-memory processing, and so forth. And you get these in the free "express" editions too, as of 2016sp1, though that is only really useful for small projects and prototypes due to other limitations in that edition (10Gb/database, max 4 CPU cores & 1.5Gb RAM put to use per instance) so you'll be paying for standard edition at least for real work.