Optimize Before Image for ProBkup online

R

RD_Man

Member
I'm running into a situation and need advice. Every 4 hours a probkup online kicks off and works perfect. The problem is, as the db/activity grow the bi takes about 3 minutes to backup and as you know it freezes activity. Is there a way to optimize the bi so that it is smaller or something else to optimize the backup. The 3 minutes of hang time are getting on my last nerve.

thx for the insight.
 
Why do you run a probkup online every 4 hours. Are you sure its probkup online or are you talking about after-imaging begin backed up.
 
I run the online so that I get a full consistent backup. I don't like having to roll in incremental's so I don't do AI. If I started to do AI's and only did a BI online every 12 hours how many AI's would I have to maintain? Or asking in a different way, can I get away with one incremental per hour? Would that be too much?
 
Why don't you use AI? You can do full backup every day and switch AI every 4 hours so you will have full backup + 5 AI every day. AI is better than incremental backup - no time required to backup (except overhead of AI writing) and you can restore database to any point in time. You should also use BIW and add more APW (and AIW if you go AI). You can also decrease size of BI block and cluster - it will probably affect overall performance causing more checkpoints but BI file will be smaller. Therefore you will have backup optimized database. You should also check your application - the fact you don't use AI suggests you have no important data in your database and that this application is not mission critical - so such big BI grow in few hours suggests that some process do something unusual.
 
1) It doesn't sound like you really understand after-imaging. This presentation might help. You should be using after-imaging. As trx says, if you aren't then you must not have any valuable data.

2) If it is taking 3 minutes to backup your BI file something is wrong. You might have an application issue and you're holding transactions open too long thus resulting in really big bi files, or you might have a really fouled up IO subsystem (for example, any form of parity based RAID). Or maybe you're writing the backup to the same disk that the bi file lives on. Or something else. Or all of the above.
 
I understand what you are saying and I'll start down the AI path. Where do I look at the current allocated % or size of the BI? I assume in promon but what am i looking for?
 
You can start by looking at "Total BI Writes", its an incremental count. Ideally you would capture it every 4 hours and then see how much space you need allocated for you AI files. All your AI files can be fixed or variable (unlike BI files).

HTH
 
I would not recommend to use fixed AI extents - I would would always use variable ones.

Heavy Regards, RealHeavyDude.
 
Another method to estimate the AI size is the following method:

Multiply the number of checkpoints that occur within a given time with the BI cluster size. AFAIK, the AI notes are almost equal the size of the BI notes. But you need to be aware that you can only see the last five checkpoints in PROMON. You could also use your average checkpoint interval during the day.


Heavy Regards, RealHeavyDude.
 
Please let us know what version of Progress/OpenEdge are you running and what OS. How many databases are you talking about here and whats their current sizes.

Planning for AI is based on the activity on the database, so if this week is not your high volume transactions or if the usage is low, you would want to factor all that in your plan. Either way backing up AI files would mean your backing up changed data versus backing up the whole database. Its always a good thing to get your basics right, some of the guys here would be available to give a few days of consultation and put you on the right track. Once you have a basic architecture setup for your databases and after imaging, all the new setups that you have would be a breeze.

HTH
 
10.1B on Windows 2k3.


Create Date: Tue Sep 02 16:20:17 2008 | Major Version: 15.0
Start Time: Thu Oct 14 04:17:43 2010 | Minor Version: --
Last Open: Thu Oct 14 04:17:43 2010 | SHM Version: V10.1 rev 71
State: open | Up Time: 106d 6h 54m 1s
BI block size: 8 KB | AI block size: 8 KB

--------------------------------------------------------------------------------

Total Blocks: 1,325,336 | Empty Blocks: 575,167
Free Blocks: 60 | RM Empty Blocks: 288
DB Block Size: 4 KB | Total DB Size: 5.06 GB
DB Block Reads: 570,145,493 | DB Block Writes: 6,865,976

--------------------------------------------------------------------------------

Last Full Backup Date: Fri Jan 28 10:00:00 2011 |
Last Incr Backup Date: | Last Incr Sequence: --

--------------------------------------------------------------------------------

Transaction Commits: 34,496,985 | Transaction Undos: 22,423
Last Transaction: 140,944,493 |

---------------------------------------------------------------------------------

BI Storage Area: Primary Recovery Area
Jan 28, 2011 10:14:21 AM

Summary details
Area Number: 3
Extent Count: 1
Area Block Size: 8 KB
Total DB Blocks: 157,008
Total Size: 1.20 GB



Extent file details
Extent # File Spec Size (DB blocks) Size (bytes) % File System Type Extend Amount (DB blocks) I/O Type
1 g:\BI\e940live.b1 157,008 1.20 GB 6 % variable 512 UNBUFIO



Extent file I/O details
Extent # Reads Writes Extends Buffered Reads Unbuffered Reads Buffered Writes Unbuffered Writes
1 98,820,942 3,665,817 1,574 -- 98,820,943 -- 3,716,186
 
Since you are on Openedge 10.1B, you may want to look at "AI File Management utility", which is a much easier option. To get more information go through "10.1B Database Administration manual" within the Openedge 10.1B documentation set "http://communities.progress.com/pcom/docs/DOC-16250".

In summary, you would activate the AI files, then you would configure the AI File management utility to manage AI extents automatically. All you have to worry about would be to backup the AI archive files.

HTH
 
You must log in or register to reply here.
Back
Top