4 ms·

KDB+ is the number 1 time-series database, no question. Per-core, it's 10 times faster than any other time-series database. Seems like an odd omission. http://

• wlesieutre 10y ago
  As noted in the scope section along with some other omissions: > Only free and open source time series databases and their features have been compared. Therefore if someone asks “have you tried Kdb+ or Informix?” the answer will be no. They are probably awesome though. Article could have been titled "Review of the Top10 FOSS Time Series Databases" since that's what it is.
  • gaius 10y ago
    The 32-bit version is free. I agree that any comparison skipping it is meaningless, it is the "standard" in this space, to the extent there is one.
    • evanpw 10y ago
      > The 32-bit version is free. Only for non-commercial use: https://kx.com/2015/09/19/32-bit-kdb-for-non-commercial-use-only/ https://kx.com/2015/09/19/32-bit-kdb-for-non-commercial-use-...
• threeseed 10y ago
  KDB+ does have some disadvantages though. Requiring a schema makes it ill-suited to many use cases since in many situations you don't know the schema ahead of time. And the biggest for me is poor Hadoop/Spark integration. JDBC is not a great choice since you have poor predicate pushdown support and it prevents say Spark from going specifically to the node with the data.
  • gaius 10y ago
    You always have a schema. The only choice you have is whether your tool helps you validate it or not.
    • zaphar 10y ago
      You always have N>0 schemas. The question of whether you have N+1 or not is governed is whether your db requires you to have a second one other than the one in your code.
• Thaxll 10y ago
  Performance is not the #1 criteria.
• pyvpx 10y ago
  I dont understand the how/why it is, though. anyone have more information?