Connecting AMPL to SQL Server

To use Microsoft SQL Server with AMPL, you need to have the ODBC driver for SQL Server installed and to have access to a database server, which could be either local or remote.



Follow the instructions in Installing the Microsoft ODBC Driver for SQL Server on Linux.


The ODBC driver for MS SQL Server often comes installed by default on modern versions of Windows. You can check if the driver is installed by running the ODBC Data Source Administrator, odbcad32.exe, and looking for SQL Server in the Drivers tab.


If the driver is missing, follow the instructions in Microsoft ODBC Driver for SQL Server: System Requirements, Installation, and Driver Files to install it.


We’ll demonstrate usage of MS SQL Server with AMPL on a small example. For this example we use the diet problem, which finds a combination of foods that satisfies certain nutritional requirements. It is described in Chapter 2 of the AMPL book.

We assume that you’ve already installed the MS SQL Server ODBC driver using the instructions above and have access to a local SQL Server database.

First download the data for the diet problem diet.xls and import it using the SQL Server Import and Export Wizard which can be run from the Start menu -> All Programs -> Microsoft SQL Server -> Import and Export Data. Skip the Welcome page, if any, by clicking Next, then choose Microsoft Excel as a Data source and specify the path to the downloaded diet.xls file in Excel file path:


Click Next and on the following page provide the connection settings for the SQL Server database you are going to use:


Click Next, select Copy data from one or more tables or views on the next page and click Next again.

Select tables Amounts, Foods and Nutrients for import and click Next:


Select Run immediately and click Finish on the next page and the one that follows.

Once import is complete, download the model file diet.mod and the script file

The script file first reads the model:

model diet.mod;

Then it defines a parameter to hold a connection string. Since the connection parameters are the same for all table declarations in our example, we avoid unnecessary duplication. In this case we specify all the connection parameters explicitly. Alternatively, you could use a DSN file name or "DSN=<dsn-name>" as a connection string.

param ConnectionStr symbolic = "DRIVER={SQL Server}; SERVER=(local);";

If you are using Linux and have chosen a driver name other than SQL Server, you will have to specify this name instead of SQL Server in the DRIVER={SQL Server} attribute in the connection string.

You can use a different version of the ODBC driver for SQL Server on Windows as well. The driver name is chosen automatically during installation on Windows, so if you are using this OS, you will have to find the driver name and specify it instead of SQL Server in the connection string. To discover the driver name on Windows, run the ODBC Data Source Administrator, odbcad32.exe. Go to the Drivers tab where all the installed drivers are listed and look for the one containing SQL Server:


A driver name containing a semicolon (;) should be surrounded with { and } in a connection string, for example:

param ConnectionStr symbolic = "DRIVER={SQL Server; version 11.0};";

Next there are several table declarations that use the ConnectionStr parameter defined previously:

table dietFoods "ODBC" (ConnectionStr) "Foods":
   FOOD <- [FOOD], cost IN, f_min IN, f_max IN,
   Buy OUT, Buy.rc ~ BuyRC OUT, {j in FOOD} Buy[j]/f_max[j] ~ BuyFrac;

table dietNutrs IN "ODBC" (ConnectionStr) "Nutrients": NUTR <- [NUTR], n_min, n_max;
table dietAmts IN "ODBC" (ConnectionStr) "Amounts": [NUTR, FOOD], amt;

Finally the script reads the data from the tables

read table dietFoods;
read table dietNutrs;
read table dietAmts;

solves the problem


and writes the solution back to the database:

write table dietFoods;

Note that the same table dietFoods is used both for input and output.

Running the script with ampl shows that data connection is working properly and the problem is easily solved:

> ampl
MINOS 5.51: optimal solution found.
13 iterations, objective 118.0594032

You can use various database tools such as SQL Server Management Studio to view the data exported to the database from the AMPL script:


SQL statements

The default identifier quote character in SQL Server is the double quotation mark ("). AMPL’s ODBC table handler detects the quote character automatically and uses it when necessary. However, user-supplied SQL statements are passed to the ODBC driver for SQL Server as is and should use the correct quotation.


table Foods 'ODBC' 'DRIVER={SQL Server};'
   'SQL=SELECT "FOOD", "cost" FROM "Foods";': [FOOD], cost;


This section lists common problems with possible solutions.

The first thing to do in case of an error is to get additional information. Add the option "verbose" to the table declaration that causes the error, for example:

table dietFoods "ODBC" (ConnectionStr) "Foods" "verbose":

Then rerun your code and you should get a more detailed error message.