Re: Can Disk mirroring/db snapshots replace the usage of taking database backups?

Sebastian, Norma J. wrote: Norma, Neil, You are both correct. I still have my head stuck in ontape land. In order to use ontape for a logical log restore following an external restore you have to have IDS 10.00 or later. Onbar could perform a logical restore following an external restore at least since 9.20 - my bad. Art S. Kagel Oninit > Art, > I must have missed something in this thread... ... > You said "Keep in mind again that if you are not running IDS 11.10 or > later you will not be able to roll forward any logical logs after a > restore" (your orig message is below). On IDS 9.4, after an onmode -c > block (external backup), I have rolled logs forward on a restore. I > must have not quite grasped what you meant, but it is late in my day and > one of my 2 brain cells has already failed over ;) > Thanks, > NJ > > > > -----Original Message----- > From: Art S. Kagel (Oninit) [mailto:art@oninit.com] > Sent: Monday, April 14, 2008 3:32 PM > To: jpierrot@chubb.com > Cc: informix-list@iiug.org; informix-list-bounces@iiug.org; Sebastian, > Norma J. > Subject: Re: Can Disk mirroring/db snapshots replace the usage of taking > database backups? > > jpierrot@chubb.com wrote: > >> *Art,* >> *The snapshots are instantaneous since it uses pointers to the disks >> as explained to me, no data transfer during a snapshot. Ideally, I >> would not want to block the DB every hour, for it will generate a lot >> of calls from the help desk. Those blocks duration must stay within >> what we are getting now with our checkpoints, 0 seconds checkpoint >> average. Anything that is noticeable to the users community is not >> good, it must be transparent. * >> >> > > Be suspicious, be very suspicious. OK, so I'm assuming that the array > is keeping an up-to-date copy on another structure somewhere, like EMC > Symmetrics, and the snapshot operates by breaking that remote mirror, > copying the remote disk off somewhere else, then reestablishing the > remote mirror. To use that kind of snapshot you will have to perform an > > 'onmode -c block' to make sure that there is a consistent image on the > disks, then request the snapshot which while not instantaneous should > take only a second, then unblock the engine with 'onmode -c unblock'. > > Keep in mind again that if you are not running IDS 11.10 or later you > will not be able to roll forward any logical logs after a restore. The > restore will be as of the moment of the 'onmode -c block' checkpoint and > > any data modified between then and the crash will be lost. Finally be > aware that until the remote mirror is completely copied (which I > guarantee is not instaneous) you will be operating without that first > line of backup. > > Art S. Kagel > Oninit > > >> *Thanks!* >> >> *jp* >> Inactive hide details for "Art S. Kagel (Oninit)" <art@oninit.com>"Art >> > > >> S. Kagel (Oninit)" <art@oninit.com> >> >> >> *"Art S. Kagel (Oninit)" <art@oninit.com>* >> >> 04/11/2008 05:14 PM >> >> >> >> To >> >> "Sebastian, Norma J." <NormaJean.Sebastian@tellabs.com> >> >> cc >> >> jpierrot@chubb.com, informix-list@iiug.org, >> informix-list-bounces@iiug.org >> >> Subject >> >> Re: Can Disk mirroring/db snapshots replace the usage oftaking >> database backups? >> >> >> >> >> Sebastian, Norma J. wrote: >> >> I have another concern. You say Falconstor will make a snapshot every >> > N > >> minutes. Do you really want to have the server blocked even once an >> hour for the several minutes it will take to make the snapshot? Or >> > are > >> they claiming that the snapshots are nearly instantaneous? If so, how >> are they doing that? By blocking the DB, breaking some mirror, then >> unblocking the DB, and making a copy of the mirror offline? Where's >> your redundancy until the mirror is reestablished? Or is it a three >> > way > >> mirror? In that case you don't even need Falconstor, you can do the >> three way mirror yourself, script the snapshots/external archives, and >> save a bundle of $$$! There have been IDS users doing that for years! >> >> Art S. Kagel >> Oninit >> >> >>> JP, >>> >>> if your assumption is right that it uses an "onmode -c block" then >>> > you > >>> would have a recoverable/rollforwardable image.... >>> IF anything goes wrong during the time the snapshot is taken.... >>> > then > >>> everybody waits till it is fixed... cuz the DB is blocked... this >>> would appear to be no different than you yourself issuing the >>> block/unblock commands... but if the auto-software (falconstor) >>> introduced between the block/unblock goes haywire for any reason... >>> well... there u are... blocked.. and so are your users.... it would >>> be wise at the least to build some type of monitoring/page-out to >>> > DBAs > >>> in the process in the event there is a snag and a manual unblock is >>> needed. (this falconstor should be able to page-out on that) >>> >>> if the software builds in a block/unblock... then it is still in >>> > your > >>> best interest to back up the DB logs via informix onbar because it >>> will be onbar, not falconstor, restoring/rolling those logs forward. >>> > > >>> (same with RMAN i would guess) >>> >>> how big is your database?... >>> how is this software going to make your life any easier? >>> what prevents you from scripting a block/split/unblock yourself >>> > today > >>> without this software?... >>> is this software being pushed on you by management? >>> >>> i only have 2 brain cells, but i can't imagine any value in adding >>> this to an IDS system. >>> do they really really support IDS or is that only lip service >>> > because > >>> they want u to buy it for oracle and/or sqlserver or something?.... >>> >>> some, if not all of your questions should be posed to falconstor.... >>> > > >>> if they truly support informix, they should be able to answer your >>> concerns. if they really want your business, they should help u set >>> it up for free on a free test server to see the value.... >>> >>> i searched their web site for informix, it came up with 2 tiny >>> references... one in legal smallprint and the other in a news >>> article... but if you search for oracle or sql server on their >>> site... .. sqlserver has 2 pages of references and oracle has 3 >>> > pages. > >>> >>> so they really support informix? put the ball in their court to >>> prove it... >>> >>> sorry... feeling critical today... >>> Norma Jean >>> >>> >>> >>> >>> >>> >>> > ------------------------------------------------------------------------ > >>> *From:* informix-list-bounces@iiug.org >>> [mailto:informix-list-bounces@iiug.org] *On Behalf Of >>> >> *jpierrot@chubb.com >> >>> *Sent:* Friday, April 11, 2008 1:42 PM >>> *To:* Kevin Cherkauer@