All about information

Selecting between proprietary and open source databases is not a trivial task

Businesses generate data that needs to be classified and organised into understandable formats. Databases rely on the concept of data models to classify facts and figures. These models logically connect, structure and organise information (for example, as tables with related data fields clubbed into rows and columns). A consolidated information store can be utilised by enterprise wide applications such as ERP and CRM.

The DB clan

Databases are classified into various categories. The most popular variety among these is the relational database.

Relational databases are based on the principles of set theory. Here the principle of presenting related data types as sets is used for classifying data. Relational databases use database management systems (DBMS) to manage queries. The combined solution is called an RDBMS (Relational Database Management System).

Pack leaders

Although other varieties such as object oriented databases (OODB) and hierarchical databases are available, RDBMS remains the most popular in terms of business deployments. An object relational database (ORDB) is yet another category that is slowly making its presence felt.

Most database vendors concentrate on RDBMS. Leading vendors in this segment include Oracle, Microsoft and Sybase. The popularity of these solutions among developers and database administrators (DBA) ensures that support (in-house/external) will never be an issue.

Open roads

If you want the flexibility to tinker with the database code, open source DBs are also available. These include products such as MySQL and PostgreSQL. Computer Associates also has the Ingres open source database. These are free databases and the only cost involved is support-related.

Many of the open source RDBMSs have wide adoption, and are considered reliable. These are good propositions for SMBs looking out for trustworthy and cost-effective databases.

The DB market

Choosing a database is a complex exercise. As with any other enterprise software deployment, the requirements from a database need to be clearly delineated before forking out the dough.

When it comes to the actual selection process, the operating system and the hardware platform to be used come first. Most SMBs typically favour Wintel/Lintel (Windows on Intel or Linux on Intel) platforms. Both offer high value at lower costs and are ideal for SMBs.

Once the platform has been decided upon, look at the features that products based on it can provide. The database should be easy to use, manage and maintain. Administration tools must be evaluated thoroughly because this is one of the most crucial features. The visual administration tools available with a database should be easy to use. This is essential for SMBs, as many of them cannot afford a dedicated, full-time DBA in-house. The services of DBAs are expensive. Therefore, it is essential to opt for a database that can be managed in-house using minimal resources. Also check for the kind of support that the vendor and its partners can provide.

Resource efficiency and supported applications are essential parameters to look for. The database must be able to run optimally on cost-effective hardware configurations. It must be able to support existing applications as well as those that may be deployed in the future. XML support in particular is useful.

The maximum number of transactions that a database can handle in a day is another important factor. This must be calculated while charting the requirements before entering the selection stage. Go in for a database that can handle a much higher number of transactions to future-proof your set-up.

Data integrity and database security are important factors. Considering that corporate databases are easy targets for blackhat hackers, built-in security components are essential. These range from access control and application roles to advanced features such as database encryption and integration with corporate identity management solution. Access auditing features are useful.

Licensing is the biggest cost component. Licenses are decided on a per server and per client basis. Be sure to check the fine print before signing on the dotted line. It is good to have a couple of spare licenses than exactly the required number. This is handy if unexpected requirements turn up. If you want to avoid licensing costs altogether, opt for an open source database. The next component adding to costs will be the amount of customisation required. As the customisation levels rise, costs increase. User training is yet another factor to consider.

Availability comes into play if you are looking at a high performance/capacity database. Many databases support availability features such as load balancing, fault tolerance and clustering that might be required for medium enterprises. However, this will come at a significant additional cost. Hence, these features do not count if there are no plans to go beyond a single database server.

If the business does not have the skill sets to handle a database implementation, it is better to seek help of an external consultant from this point onwards. The consultant should have sufficient and varied experience in implementing, administrating and supporting databases. It is a good idea to ask for a list of his existing clients. Speaking to these clients will help in understanding whether he is the right person.

When negotiating a deal with the consultant, check about ongoing support. If this aspect has to be handed over to another consultant, it might take long for the new consultant to get used to the database logic and the data models used.