Databases Reference
In-Depth Information
releases is dropped. That means you will not be able to get help if something
stops working. In addition, as bugs and problems are identified, the solution is
usually offered in a newer version of the product. If you are unwilling to move
forward, you have to live with these bugs. Most organizations begin to panic
when support will be dropped. The path to upgrade is then under pressure
and is based upon the vendor support timeline, rather than a schedule that is
convenient or affordable for your organization.
When support is dropped, it is usually after several releases, so getting
your environment up to current standards often means jumping ahead two or
three releases. This can be a big change; and there can be major differences in
supported configurations, how to install the software, and how to use it. These
changes are often less dramatic when taken one release at a time. It is usually
less expensive in the long run to incur upgrade costs as each release becomes
available. Costs associated with upgrades may include software, retraining of
IT and business staff, and conversion of existing programs or reports. There
may also be a delay of new development when resources are focused on
upgrading the software. The costs and time required is typically less when
moving up one release at a time compared to jumping several releases at once.
It is worth the investment to keep your data warehouse up to date with
your technology. Once everything is current, make a commitment to keep it
that way. This often means a steady stream of new product releases (some for
maintenance and some for new features). Installation of these new product
releases can often be bundled with other maintenance and enhancements that
are made to your data warehouse environment.
Monitoring Performance and Capacity Planning
Once the data warehouse is launched, it takes time for its use to become a
regular part of the day-to-day business. Data volumes will grow as you keep
loading data. The number of users will increase as more people get comfortable
with the environment. The number of queries and the complexity of queries
will also grow over time. All of these factors can lead to performance or
capacity issues.
The performance of the data warehouse must be monitored in two separate
areas. First, the performance of the ETL system must be tracked and monitored.
There is often a processing window of time that must be met to deliver the data
in a timely manner. Often, this window is specified as part of a series of service
objectives that are defined and agreed to by business and IT management. The
performance of the ETL system is not usually observed by the users of the data
warehouse unless the data is not ready on time. IT organizations are typically
already equipped with the tools necessary to monitor performance of the ETL
system.
