Archives of the TeradataForum

Message Posted: Tue, 02 Nov 2004 @ 20:01:42 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

<-- Anonymously Posted: Tuesday, November 02, 2004 10:51 -->

Hi all,

I have a couple of archive questions and I would know what is your experience in the matter.

I am working on an 8-node 5250 system. We use ASF2 to backup our data (we are running v2r4), and we have 6 drives attached (3 nodes with two drives each). We do cluster archives the our whole databases. The amount of GB archived increases by 3GB/day more or less. The distribution of data is very good, so every job has to backup 1/6 of the data.

In the last months we are experiencing a big problem with backup, that I'll try to explain.

It sometimes happens to last several hours, and this is because there is a single job that happens to be very slow. In these cases its throughput is very low (near 1GB/h or less). These huge slowdowns happen even when our system seems to be unloaded; if we look for something executing in the system using PMON, we aren't able to find anything that could justify what is happening.

I wonder if there could be any activities in the system that we aren't able to monitor at all; I wonder if these "ghost activities" could be responsible for this situation too.

We use PMON as well as schmon for monitoring partitions and sar on UNIX nodes.

Moreover in these cases (when it lasts several hours) the situation seems to be very strange; in fact on a single node there is one slow job, while the other is ok (throughput near the maximum expected).

How is it possible? I would expect that two jobs on the same node would share the same resources, and so would have the same throughput, but perhaps it is a wrong assumption....

Has anyone experienced the same problems with archive and may throw light on these arguments?

May anyone throw light on how does backup work at its very heart?

Thank you everyone

	<Prev	Next>		<<First	<Prev	Next>	Last>>