From: Alexander.Kuznetsov@marshmc.com (Alexander Kuznetsov) Newsgroups: comp.databases.oracle Subject: Re: Where to publish an article? Date: 11 Nov 2002 12:42:21 -0800 Organization: http://groups.google.com/ Lines: 17 Message-ID: References: <548b9514.0211110759.7ec3c941@posting.google.com> NNTP-Posting-Host: 4.20.74.62 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit X-Trace: posting.google.com 1037047342 22108 127.0.0.1 (11 Nov 2002 20:42:22 GMT) X-Complaints-To: groups-abuse@google.com NNTP-Posting-Date: 11 Nov 2002 20:42:22 GMT > > Perhaps I did not quite understand, but how is your process more > efficient than the Oracle Sequence Numbers. > if the numbers are consumer in the middleware, you save a lot: a rounf trip across network for every number > Also instead of defining version_id as number, define it as date or > timestamp and set it to sysdate or systimestamp and using that in the > where clause should eleminate the 'nasty problem' > No way. On a powerful box there may be several updates with the same timestamp value. With DB2, timestamp are accurate to microseconds (0.000001 sec) and still attempts to have a unique index on MODIFIED_AT _DO_ fail on fast enough machines