Hi.


When we create a sequential dump file from Baan, the first two lines created contain the following data

2007-01-12 15:35:05.890
#$!pre-dump!$#00017cv


What is the second line ?

Basically, I am trying to create a customized session to create sequential dump file.

Dear vineetu1,

looks like the separator in hexadecimal:

7c = | (Pipe sign). (is 124 in Decimal system).

If you try ";" the it should be 3b.

Please take a look here with detailed information:
http://www.baanboard.com/baanboard/showthread.php?t=8855&highlight=%23%24%21pre-dump%21%24%23

Regards,

Kozure

Hi Vineetu,

Search here..this is discussed earlier.


May I know, whats the reason you are customising seq. dump session?

You can use bdbpre on prompt.

regards,

-Suhas

To speedup Backup restore.

The size of sequential dump is around 18 GB. It takes approx 44 hrs to restore the data in the Backup server.

I find that around 80% (15 GB) of the size of these dumps is by a few big tables ..i.e. tfgld106, tfgld418, tfgld410, tdinv700.

Basically I wanted to device a way to create a sequential dump for selected records (i.e. records which have changed). As this is not supported by Baan as confirmed by them hence I have decided the following :

I intend to create a customized session which will create a sequential dump for the above mentioned tables for only few records (i.e. latest records).

Records of current year. (Old year records are not changing hence no need to again restore them)

While restoring I will chose the option of over writing the existing records.

With the above, I will be able to reduce the restoration time by upto 70%

Let me know your views.........

44 Hrs. are too much. You really have to look into Fine and Tuning. May be more on Hardware, Disk Layouts/Contention, Application Tuning and Ref. tables.

Vineetu, I am sorry to disagree with your way, reliability and test phases. More curious, How you are tracking old data changes and current financial year data?

If BaaN doesnt support incremental backup, do you look into Informix Utilities? or Have you checked NTBACKUP or any supported backup hardware?

regards,

-Suhas

Hi Suhas,

Yes performance tuning through proper disk layouts, separating indexes and data, will help to some extent, but considering hardware limitations due to cost factors, the best way to drastically decrease the restoration time is what I have thought off.

Now how do we track the back dated entries, this would involve some manual intervention and would not be totally automated. But the frequency of intervention required would be very less.

For example, the live server has data from 2001 onwards (pls don't ask why we have not done archiving, there are valid reasons for that). All the fiscal periods till Dec 2006 are hard closed. Hence there would not be any transactions in year 2006 and before. To be extra safe if I take seq dump from 2006 onwards or mid 2006 onwards, still it is a safe thing.

As the no of Big tables are only 4-5, hence these can be monitored individually without much effort.

So, in our case, this solution fits.

With regards to OS level replication and other issues, one of the reasons why it is not possible is because the live server and backup server h/w are not identical.

Yes, will have to look into Informix level tools, but it will take more time as there is quite less information (knowledge base) regarding Informix as compared to Oracle. Oracle is the most popular database, most widely used.

All the best.

Keep posting your experiences.

regards,

-Suhas

Are there any options availabe with bdbpre create dumps based on certain value of table fields or Index ??

Hi vineetu1,

no, you can't specify a range of records to be dumped with bdbpre.
But I guess that bdbpre isn't your problem (concerning runtime), but bdbpost is.
So what you could do (instead of writing a Baan program to dump certain records only) is to filter the seq.dump file with (for instance) awk.

hth

Norbert