Networked Database speed issue...

Question

I apologize... I am kind of new to programming and have a limited scope of understanding when it comes to databases and sql queries.  I was given the task of creating an application to perform a sorting algorithm and implemented it using LabVIEW 2013, the database toolkit, and a MS Access 2007 database.  The application I wrote worked great when it was first deployed and now that it has been operational for about 2 months, it has slowed significantly. I did some testing and I think I have isolated the latency to network communication issues.  
I believe it is network speed/traffic related because I copied the database locally and sorting was only about 2s - the networked copy takes about 14s!!! Also, the number of users had no impact on this number either - I created a copy of the database, renamed it, saved it to the same folder, pointed my application to that file instead and it was still 14s...  The only noticeable improvement was using a local copy, which will not work for this scenario.  
I spoke to a few people about it and someone suggested changing the cursor location from server based to client based (http://msdn.microsoft.com/en-us/library/windows/desktop/ms676564(v=vs.85).aspx).  Only thing is, I cannot figure out which LabVIEW is using or how to change it... Is it changeable?  If not, do you have any other suggestions for speeding it up?  It is unlikely I will get much support from our IT department, so any improvement suggestions will have to be recommendations for change to my application.

Answer 1

I suppose I should have been more clear when I said sort...lol.  The application is used to find a PO for a particular device (sorting to a PO).  There is only one sort operation (a->z) performed in any of my sql queries (see below)
I was concerned about using the database as well - mostly because I'm not as experienced using them...  The original implementation of the application used text files and flags to prevent data collisions, but we have about 10 people sorting at one time and it was creating a huge lag time.
The network location of the database (we believe) is probably what is causing most of the issue.  Our IT dept keeps things very locked down and there weren't many choices for implementation methods.  I was forced to use a shared folder on the network to store the database and this shared folder is accessible to everyone in our facility.  Be believe the general traffic to this server is probably most of the issue we are having.  
when the version 2 sort tool was deployed (the one using the database instead of text files) it took just a few seconds to return a result - on the order of 3-5s.  There are now over 150,000 entries in the database, so I'm sure that too is contributing to the lag.  
The database has several tables:
(1) Log Table: used strictly for logging the device sort data,
(2) Liquidation Table: lists all of the possible sort locations and the device requirements for each PO,
(3) Activity Log - used for capturing devices which shouldn't be sorted or other device condition based exceptions,  
(4) Allocation table - for each model it lists whether the device should be sorted to a PO based on the priority of the PO or an allocation percentage, and
(5) Requirements table - only 3 entries.  Used to control specific behavior of the sort.
I have several SQL queries in the code that hit this database...  
1. Check the requirements table.  The flags reside in the table to (1) stop the sorters from being able to sort devices temporarily and (2) a flag to force all devices to be cleared and/or graded before they can be sorted.  
SELECT Requirements.Requirement, Requirements.Enable FROM Requirements
WHERE (((Requirements.[Start])<Now()) AND ((Requirements.[End])>Now()));
2. Check the log to see if the device was already sorted - if it is, it flags that log entry so it is not counted against a PO and can be resorted.  
SELECT ID  FROM LQD_Log
WHERE PO_ID > -2 AND ITEM_ID ='[ITEM ID]'
If not null, a case statement updates the PO_ID for that log entry to a value that will not count against any open POs.  
3. The next query is a little more complicated.  Depending on the device condition, several case statements build the SQL query string to find a PO the device can be sorted to.  Since there could be multiple POs returned, I have it return the POs (at most 5 or so) sorted by priority.  (there are 13 elements for each PO returned).
4.  The next query checks the allocation table to see if the that model is to be sorted based on priority of the PO or by an allocation percentage.  
SELECT Percent  FROM Allocation_Table
WHERE CNN ='[COLOR NEUTRAL NAME OF DEVICE]'
A PO is selected in my LabVIEW code based on the POs returned and the allocation method. 
The PO (Sort location) is displayed to the user.  they scan the sort location and then I write the information to the log in the database - Pertinent device characteristics related to the sort decision, PO Information, user information and scanned location (21 elements).  
There is very little which can be done about the location of the file... it has to be in a shared location, and it has to be able to support multiple user access.  One of the programmers here recommended using an Access database for this reason and many others.  I could possibly lock it down with flags again, but I'm afraid that would only contribute to the problem - I did test a copy of the database to the same network location and I saw the same types of delays, so I really do not believe it is related to multiple users accessing the file.