Re: How do Banking Systems Work

From: Jonathan Lewis <Jonathan_at_jlcomp.demon.co.uk>
Date: 1995/12/19
Message-ID: <819408499snz_at_jlcomp.demon.co.uk>#1/1


In article <4an77a$lfr_at_rcp6.elan.af.mil> franks_at_qnet.com "frank surewood" writes:

: Example, how does my ATM
: respond so quickly when the select must be in the same scale as my
: application to find my records? Is the answer big and better hardware?
:

Design, and the efficiency of the balanced B-tree.

A query for a bank balance is probably against a summary table that is either kept up to date in real time, or batch updated overnight.

Your query is then by primary key.

Simple arithmetic suggests that a 75% efficient index on a 4K blocksize with a small (12 bytes) key will be able to pick one row out of 2,000,000,000 in less than half a second even when run on a fairly small machine.

The big bug-bear on large databases is not the size of the tables, it is the simultaneous usage (lots of people want a couple of rows each) or number-crunching queries (a few people want a lot of rows each).

Row throughput, not inherent database size, is in my experience the controlling factor on an Oracle system,.

-- 
Jonathan Lewis
Received on Tue Dec 19 1995 - 00:00:00 CET

Original text of this message