Setting data standards
One of the critical aspects of data administration is to ensure that standards for metadata are in place – for example, what metadata is to be kept for different underlying ‘data types’. Different details are kept about structured tabular data than for other areas. ‘Ownership’ is a critical item of this metadata, some sort of unique identifier is another, a description in business meaningful terms another, and a format might be another. The custodian or steward, someone in the IT department who takes responsibility for the day-to-day management of the data, is also recorded.
Another benefit of a Data Management process would be in the field of reference data. Certain types of data, such as postcodes or names of countries, may be needed across a variety of systems and need to be consistent. It is part of the responsibility of data administration to manage reference data on behalf of the whole business, and to make sure that the same reference data is used by all systems in the organization.
Standards for naming must be in place; so, for example, if a new type of data is requested in a new service, then there is a need to use names that meet these standards. An example standard might be ‘all capitals, no underlining and no abbreviations’.
Data administration can assist the service developer by making sure responsibilities for data ownership are taken seriously by the business and by the IT department. One of the most successful ways of doing this is to get the business and the IT department to sign up to a data charter – a set of procedural standards and guidance for the careful management of data in the organization, by adherence to corporately defined standards. Responsibilities of a data owner are often defined here and may include:
Data migration is an issue where a new service is replacing a number of (possibly just one) existing services, and it’s necessary to carry across, into the new service, good-quality data from the existing systems and services. There are two types of data migration of interest to projects here: one is the data migration into data warehouses etc., for business intelligence/analytics purposes; the other is data migration to a new transactional, operational service. In both cases it will be beneficial if data migration standards, procedures and processes are laid down by Data Management. Data migration tools may have already been purchased on behalf of the organization by the Data Management team. Without this support, it’s very easy to underestimate the amount of effort that’s required, particularly if data consolidation and cleaning has to take place between multiple source systems, and the quality of the existing services’ data is known to be questionable.
One area where technology has moved on very rapidly is in the area of storage of data. There is a need to consider different storage media – for example, optical storage – and be aware of the size and cost implications associated with this. The main reason for understanding the developments in this area is that they make possible many types of data management areas that were considered too expensive before. For example, to store real-time video, which uses an enormous bandwidth, has, until the last two to three years, been regarded as too expensive. The same is true of the scanning of large numbers of paper documents, particularly where those documents are not text-based but contain detailed diagrams or pictures. Understanding technology developments with regard to electronic storage of data is critical to understanding the opportunities for the business to exploit the information resource effectively by making the best use of new technology.
It is also very important to work with Data Management on effective measures for data capture. The aim here is to capture data as quickly and accurately as possible. There is a need to ensure that the data capture processes require the minimum amount of keying, and exploit the advantages that graphical user interfaces provide in terms of minimizing the number of keystrokes needed, also decreasing the opportunity for errors during data capture. It is reasonable to expect that the Data Management process has standards for, and can provide expertise on, effective methods of data capture in various environments, including ‘non-structured’ data capture using mechanisms such as scanning.