I have been using Sesame for a number of years now, first as a way of up-dating old Q&A databases and then to create many production and quality recording data. I have finaly got around to the last Q&A one that was programed about 20 years ago!

Our problem is that it is a serial numbering and repair database that has about 520,000 serial numbers in that the repair database looks up for origional information using the serial number as the shared key. This can take between 7 and 12 seconds which used to be instant in Q&A due to it's "Speedy" fields.

We are running the Sesame server on a HP Windows 2012 instance with 4 Gig of RAM and 2 CPUs, which does not seem to break a sweat during the XLookup process, 4% load at best. The Network is all Gigabite speeds.

I will increase the server to 8 Gig and 4 CPUs over the weekend to see if there is any improvement.

Looking through the Sesame manual it says it does not have the Q&A speedy fields because "In Sesame, fields are indexed as needed without user intervention."

Is there a way of forcing this or does anyone know anything about this process so that I can capitalise on it?

Am I just using too many records in one Data-bases?

Any views or advice would be helpful.

Thanks in advance 

Thankyou!

That was just what I was looking for. Back to instant lookups. 

The Cow wrote on Nov 26^th, 2014 at 9:06pm:


Does this help with @XListValues? I'm not seeing any difference with it ON or OFF.

Duh! Sorry about that. I didn't think it through very well, did I?

Currently 13,094 records. The db file is 13.1 MB, and the dat file is 178 MB.

It takes about 8 seconds on my i7 computer, but it takes 13 to 18 seconds on the client's computers.

Someone in-house set up a combo box that populates with all the entries each time they open the retrieve spec. I may implement a different method where I filter the list down based on some input by the user, rather than fill it with every single entry in the entire database.

I suppose the question comes down to, is the slowdown in retrieving 13,000 values and sending them across a network, or in trying to place and manage 13,000 in a combo box, or both.

Have you timed the retrieve separately from the populating of the combo box?

You might be interested to know that Sesame3 has a "table of contents" window that runs off of the sort spec, that might be awfully useful in similar circumstance - basically a way to retrieve and navigate that does not rely on the user knowing the data in advance.