Re: Question on Structuring Product Attributes

From: -CELKO- <jcelko212_at_earthlink.net>
Date: Tue, 18 Oct 2011 09:36:10 -0700 (PDT)
Message-ID: <1204d679-a881-4882-ac15-45e522edf1e5_at_hv4g2000vbb.googlegroups.com>


>> Actually, Standard SQL does not support indexes at all.  They are an implementation detail that actual SQL DBMS use, but they are completely outside the scope of the SQL standard. <<

You already read my short history of X3H2 and CREATE INDEX, so I will not repeat it. Back in the ealry days, we worried about how to do the indexes. B-Tree in its mutations were favored. Oracle briefly had a ersion that carried the prefix of CHAR(n) and VARCHAR(n) strings at different level; it had too much ovehead on INSERT and UPDATE.

Today, I would look at hashing and brute force scans to enforce unique constraints. In a large DB, I have to traverse several levels in an index, but a good hash is one probe no matter how big the table is.

If I have a massively parallel procesor and SSD, I can scan the SSD at processor speeds; everything is effectively in Main Storage. Why bother with indexing and all the troulbe to maintain it? Received on Tue Oct 18 2011 - 18:36:10 CEST

Original text of this message