DB2 9.7 FP 3a /Win 2008 R2

our DB size is around 550GB, contain around 25 tables and around 100 indexes.
each table contains average 60 million rows.
This is a HADR enviornment, the system daily produces around 50 transaction logs each 10MB size.
Daily we load data into the DB , around 15K rows (include image -each image 2-4 KB size)

I am going to do a bulk load (which will effect most tables) and expecting around 250,000 rows including images .

The question is :--
1. How can I workout that howmany transaction logs will be produced during the load
2. As this is a HADR enviornment, will this effect log shipping? bandwidth is enouth?
(current banwidth is 4mbps between primary and standby).
3. Update Statistics is required immediatley?

4. Also any other concern?

Please advise.
Thanks, valan

hi,
anyone can give some ideas please?

Also, i mentioned the bulkload. but i am not going to use the import or LOAD.
going to use the standard insert statement.

Thx.

1) We cannot give you a specific amount. There are a lot of things that determine how many log files will be generated. I suggest you test it out in a test environment and see.

2) Of course it will affect log shipping. You are adding work load to the system. Whether the bandwidth is enough depends on the outcome of how many log files are being created.

3) You are adding less than 0.5% rows each time. Stats may not be needed each time the bulk load is performed. It depends on the kind of data you are loading.

Andy

Thanks Andy.
I can not try this on the test system as the DB is very small and not enough space.

Is it advisable to suspend the HADR (Deactivate the DB on standby node), once the load has been completed then resume the HADR?

I do not think so. But I am no expert on HADR.

Andy

I don't think you need to worry about HADR if you are only inserting 250,000 rows, assuming your log buffer is set correctly. But I am not sure what you mean by bulk load.

You should be able to test this on some system using a subset of the rows, and extrapolate to a larger number of rows.

Is the bulk running during downtime or not? If it is not in downtime you must not deactivate your HADR either. You have HADR to guarantee db-uptime so why take extra risks.
On the other hand: when you do run during downtime, why use INSERT's? LOAD's are much faster....

do not forget copy yes when using load and hadr - and logging of index create