The Yard Relational Database System

I am a relative newcomer to Linux, having used it for about three years. In that time, I have been impressed by the creativeness of the Linux community, and its dedication to Linux.

Having caught the fever myself, I find myself telling my acquaintances in business about Linux and what it can do for them. Some are willing to listen and will use Linux for some narrowly defined tasks, such as a firewall. Most, however, are reluctant to use Linux as a part of their core operation. This is due in part to a reluctance to entrust their businesses to a free operating system—“free” somehow implying cheap or inferior. But a large part of the reluctance, I've found, is due to a seeming lack of applications that run under Linux.

In particular, the fact that none of the “big three” relational database packages are available in Linux editions is a serious problem for many business people. As businesses come to depend upon relational databases for managing their day-to-day operation, having a robust relational database management system (RDBMS) available is crucial to selling an operating system to business people.

Fortunately, a solution to this problem exists if one is willing to consider software written somewhere other than northern California. Many companies in Canada and Europe now offer their products with Linux support, and some of those products offer much of the functionality of better known products at a fraction of the price. One such product is the Yard RDBMS.

Yard is an RDBMS package that is published by Yard GmbH of Cologne, Germany. The Yard package includes the following:

Yard is a fully featured RDBMS. Its SQL engine includes the following features:

Yard uses a standard client/server interface: the SQL engine services requests from client processes that are running either locally or on networked systems. Local requests are received via pipes; network requests are received via sockets. Client processes can use a number of protocols to exchange data with the engine: ODBC, JDBC, ESQL-C, etc.

When you install Yard, you must define one or more database systems (DBS). Each DBS has its own shared-memory segment that is used by every user who is working with that DBS, and one or more database spaces, each of which is a chunk of disk that holds data. Yard's method of managing system resources closely resembles that of major commercial relational database packages; if you are familiar with Informix OnLine, you will feel at home working with Yard.

Each of Yard's database spaces is a statically allocated portion of disk. Usually, a database space is a part of a raw disk partition—one that does not have a file system on it. A database space can also be a file, although this is discouraged.

The system administrator must use a fairly complex formula to compute just how much disk space to allocate to a given database space. This formula includes the number and size of the physical and logical logs, the number of users, the number of tables and the estimated extent of each table.

The administrator can dictate the database space used by individual tables within a database. By estimating the frequency with which individual tables will have to be accessed, the administrator can balance disk I/O across all of the devices that hold a given database and thus ensure transactions are processed as quickly as possible.

As you can see, Yard is a serious package. To see it at its best requires serious hardware—preferably a machine with lots of memory, cycles to burn and multiple SCSI disks. It also requires an administrator who knows databases, UNIX and hardware and who can devote a significant portion of her time to monitoring and managing the database.