In the "good old mainframe days" all company critical data was stored centrally, in large computer centers manned by specialists. But in the 90’s virtually all corporations have decentralized storage and maintenance of databases to subsidiaries and smaller business units. However, database maintenance knowledge has not followed suit: On the contrary, it is completely missing in many companies. This poses a serious business threat to many middle-sized companies, a threat that sofar largely has been ignored.

The "down-sizing trend" of the databases has lead to increasing vulnerability and unnecessary large business risks considering the importance of the systems; Customer registers, invoice systems, accounting systems, systems for wages and salaries, supplier registers etc are utterly critical for every corporation, and a loss of data normally implies grave consequenses to the operations of the company. What do you do if you lose your entire customer register just before a large marketing campaign? Or drop your wages system so that the wages cannot be paid on time?

The problem is that you often do not notice the deficiencies of your databases until it is too late, i. e. when they are already destroyed. Despite the ostentatious advertisements it is a fact that the major database maintenance programs on the market – MS SQL Server, Oracle, Sybase – have a number of weaknesses that, if the worst comes to the worst, can threaten the databases maintained by them. (This must surely come as a surprise to many less experienced network administrators, who believe that they have insured themselves against loss of data by buying a well-known database management system.)

Furthermore, no system is better than its "boss": if you mismanage your databases they will inevitably go corrupt, regardless of which program you use.

Therefore many corporations must ponder over how they really maintain their databases. Some of the measures to be taken include:

• To overhaul your maintenance routines – if you have any! – and to assure yourself of adequate competence in database administration, either in-house or external.
• To regularly (preferably on a yearly basis) conduct "disaster drills", during which you simulate a database crash. This way you can detect the weaknesses of your maintenance routines.
• To use any of the new administration programs that facilitate maintenance and improve the performance of the most common database management systems.
• Most important of all is to continuously – and often! – do backup of all data since the database maintenance programs do not perform that function with automation (another fact that must come as a surprise to many database managers).

It is possible to avoid time-consuming – and expensive! – "surprises" in the form of destroyed databases and loss of data. But it is of no use to "buy a fire-alarm when the house is already on fire" – you have to be prepared. And this is not only a matter for the computer department: The data that is stored and maintained locally is so company critical that routines for administration of databases must be handled as a strategic issue for the management in all corporations.

Dagens Industri 8/18/98
Computer Sweden 11/06/98
Nätverk & Kommunikation 19/1998