Slashdot Mirror
Digg Says Yes To NoSQL Cassandra DB, Bye To MySQL
donadony writes "After twitter, now it's Digg who's decided to replace MySQL and most of their infrastructure components and move away from LAMP to another architecture called NoSQL that is based in Cassandra, an open source project that develops a highly scalable second-generation distributed database. Cassandra was open sourced by Facebook in 2008 and is licensed under the Apache License. The reason for this move, as explained by Digg, is the increasing difficulty of building a high-performance, write-intensive application on a data set that is growing quickly, with no end in sight. This growth has forced them into horizontal and vertical partitioning strategies that have eliminated most of the value of a relational database, while still incurring all the overhead."
Postgres, for people who care about their data.
Go with PostgreSQL. Reliable, standards-compliant, fast.
___
If you think big enough, you'll never have to do it.
I have worked with large PostgreSQL databases (150GB or so) and really, Postgres isn't a solution. You run into issues anyway when some of your tables contain millions or even billions of rows. At that stage things like vacuuming or altering the schema start to become damn near impossible, and even querying starts to become a bottleneck.
Now how do you scale that if your database is still growing? Postgres doesn't have a decent clustering solution that I know of, so your options are either to roll your own, or to scale vertically. Both of those are expensive options.
Based on my experience, I don't think that relational databases are appropriate for really large databases, and at present the only realistic option is horizontal scaling which is a lot easier with things like Cassandra or MongoDB.
Free posters and articles for business analysts and project managers
Teradata and the other big relational db products (vertical, greenplum, etc) are all _analytical_ databases, designed for small amounts of complex queries, where adding new data to the system takes minutes if not hours. They are completely unsuitable for running a live application against.