Re: Theory of Timeseries extensions to SQL and database

From: Jan Hidders <hidders_at_REMOVE.THIS.uia.ua.ac.be>
Date: 30 Oct 2002 11:06:28 +0100
Message-ID: <3dbfaf24$1_at_news.uia.ac.be>


David wrote:
>
>Click on http://www.sandtechnology.co.uk/products/index.html then Library
>then Nucleus Server Whitepaper. It is kind of hidden away.

Hidden away in the UK. :-) Found it, thanks.

>Any idea how a system that replaced the bit array with a hashtable might
>perform?

Hard to say. Since I don't know their compression technique I cannot say under which circumstances it performs good or bad. My gut feeling is that if you split the original table in binary tables, replace the actual values with references to them and introduce the proper indices (hash table or otherwise) that will be a pretty good general solution, especially if some of these tables/indices can be kept in main memory.

  • Jan Hidders
Received on Wed Oct 30 2002 - 11:06:28 CET

Original text of this message