DSXchange

Honestly, I don't know squat about ZFS and I really wouldn't worry about which filesystem you are using, rather worry about it being configured correctly and optimized for the type of I/O you are seeing.

Then it would seem you need to renew your Premium membership.

In addition to that, what database are we talking about here and what kind of perceived issues are you seeing? Generally speaking, one does not "research the ODBC stage" looking for speed improvements, especially over a native stage.

Yes. Filter any 'empty' records post-pivot.

jwiles wrote:If I hadn't already done my thinking for the month I might have a suspicion about it...

Dang, James. You need to more careful rationing your thinks.

So you didn't actually migrate 'the server', correct? To me, that would mean some sort of backup/restore whole-blood copy of the file structures but that wouldn't be prudent unless so many things were identical across environments, including the operating system. I'm assuming your vendor actually mi...

Yes, your file system can make a huge difference. Keep working with your Administrator to improve that.

There is no difference.

That 'freezing' is to be expected while records are flushing to disk. First suggestion - don't use writing caching. Second suggestion - create the hashed file with a large minimum modulus. Use the Hashed File Calculator to get an idea of what the value should be.

That is odd... how exactly was it 'migrated' and from where?

Of course, just trying to point out something that it seems many people overlook.

Exactly, you must run the ulimit command from inside the job's environment, not just simply at the command line. If that's what you did.

Welcome.

First suggestion is always to search. This has been addressed here multiple times, please do an exact search for "Variable previously undefined" and see if any of the previous discussions help.

Link ordering would ensure that the 'master' record is sent to its target first before the child record would begin its journey. If this was a Server job you could use a single target stage and thus keep all of this work in a transaction or unit of work which can cut down on the shenanigans needed. ...

I agree.

And Eric is absolutely correct as to your issue. The way you have them now they just get passed in as strings rather than being resolved as parameters.