HP3000-L Archives

March 1996, Week 2

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Jerry Fochtman <[log in to unmask]>
Reply To:
Jerry Fochtman <[log in to unmask]>
Date:
Tue, 12 Mar 1996 12:21:04 -0700
Content-Type:
text/plain
Parts/Attachments:
text/plain (31 lines)
At 12:09 PM 3/12/96 -0500, Randy Smith wrote:
>I had read somewhere that Turbostore on 5.5 was going to allow true on-line,
>zero down-time backups.  Is that still scheduled?
 
Previously this was announced by HP.  I don't believe all the details of 5.5
have been announced by HP but perhaps someone (Jon?) from HP could comment
if this is still on target....
 
>One of our analysts would like to know what the mechanism for that is;
>especially dealing with and maintaining the integrity Image/SQL databases,
>and possibly Oracle.  Anyone know?
 
A new internal procedure called 'DBQuiesce' was implemented by the HP IMAGE
Lab team to help facilitate true on-line backup.  It basically allows the
store process to 'block' further access to a database, like a 'superlock'.
Transactions which are in-progress are allowed to complete while any that
are started after the 'DBQuiesce' are suspended.  Once all the in-progress
transactions have completed, the backup product can proceed to copy the
database.  After the copying is finished, the backup process DBUNQuiesce's
the database, at which point those processes that were 'blocked' simply
continue.  Basically the processes do not have to log-off.
 
There may be situations whereby in more sophisicated application
environments when multiple IMAGE databases are involved this process cannot
guarentee full logical data integrety.  However, this may be no different
than what is available by the other 3rd party tools.  You'll certainly want
to review your specific situation.
 
Oracle DBMSs are not addressed by this process.
-- Jerry

ATOM RSS1 RSS2