There are some things about this latest version of Oracle that some people do not like. I agree with Oracle’s decisions, but want to mention some of these complaints here:

•  Oracle8 is not 100% CORBA compliant. CORBA stands for the Common Object Request Broker Architecture. It is a technology standard that allows applications to use functions and data (objects) from within other applications. A competing standard is the DCOM (Distributed Common Object Model), promoted by Microsoft. I mention this because this may be your first introduction to shared objects if you have ever included a link to a spreadsheet in the middle of your word processing document. The code to read and process the spreadsheet data is not duplicated in the word processing code. Instead, it is accessed from the spreadsheet software. Many purists have objected to the fact that Oracle did not include full CORBA support as part of Oracle8. The CORBA technology was still being finalized as Oracle8 was being developed. It is much easier for a Silicon Valley startup to jump on the bandwagon of potentially unstable technology than it is for a company such as Oracle that is being used as the basis for a large number of business data processing centers. Oracle will incorporate compliance with CORBA in a later revision when both CORBA and Oracle are ready for this new technology.

•  Java is not a native procedural language for the database. PL/SQL still reigns as the internal programming language. Again, there are plans to include Java as an internal programming option for the Oracle database management system. However, the Java language still continues to evolve at a somewhat dizzying pace. When Java and Oracle are ready, it will be an option for internal programming. Until then, you still have the option of writing Java applications to access the database (via J/SQL) that are stored external to the database.

The latest release of the Oracle database management system is definitely something worthy of your consideration. While many leading-edge enthusiasts salivate at the object-oriented technologies and possibilities for Web cartridges, the Oracle database managment system still provides support for all of the traditional databases that were developed for Oracle version 7. You can run an application and database developed for Oracle7 without any conversion. When you do move to Oracle8, you will be able to take advantage of many of the “under the hood” improvements for speed and scalability that have been built into Oracle8. When a new customer asks whether they should buy Oracle8 or start with Oracle version 7, I recommend going right to Oracle8 and using whatever features they need and are comfortable with. It is everything Oracle version 7 was, and more.

Major Improvements

First, let me discuss the enhancements to the Oracle technologies that were available in version 7.3 that make Oracle8 even better. Most of these are not directly observable by the average user, developer, or even DBA. However, in most of the actual business uses of Oracle, they are probably far more important than the new features that will be discussed later. The key enhancements to existing technology in Oracle8 include the following:

•  Parallel updates, inserts, and deletes. Oracle has supported parallel queries since Oracle 7.1. In the past, updates, inserts, and deletes have been serial in nature (one process works to complete the given task). With partitioned tables and indexes, Oracle8 can now assign a number of processes to complete the insert, update, or delete process, thereby speeding up these operations. This can become significant in such tasks as data warehouse loads.

•  Incremental backups. Previous versions of Oracle have relied on complete backup strategies. You backed up an entire tablespace or database using the cold and warm backup techniques. You performed exports of an entire table or set of tables. This is easy on relatively small databases (a few tens of gigabytes). However, for very large data warehouses, this can become impossible (there is not enough time in a day to complete the backups). The incremental backups allow you to perform multiple levels of backup, capturing only the transactions that have occurred since the last time this level of backup was performed. A level is a reference point starting with 0 (complete backup). Higher numbers indicate that these backups are performed more frequently and usually back up less information than lower numbers.

•  DBLink concentration. Supporting a large number of users (thousands or tens of thousands) can become quite a challenge for a database server. One solution is to implement multiple tiers of processing. One server, for example, can perform most of the business logic processing while another performs the database retrieval functions. DBLink concentration allows multiple user connections to share one physical (network interface) and operating system connection to link to other databases.

•  Connection pooling. Connection pooling allows large number of users to connect to a database. As opposed to allocating memory and processes for each individual user connection, the system uses a pool of connections that all users share. Only the active users consume a resource.

•  Improved parallel server. For reliability, Oracle implemented the parallel server in version 7. Under a parallel server, you have multiple computers accessing a single database on a shared set of disk drives (where the hardware architecture permits such sharing). In Oracle8, this functionality has been improved to reduce downtime during the failover process. Also, the performance has been improved by minimizing the overhead tasks associated with the parallel server.

•  Improved replication. Replication is used to duplicate a set of tables from one database to another. In Oracle 7, the replication was based on row IDs and used a series of triggers to implement the process. In Oracle8, the replication function has been improved in several ways. First, parallel propagation is now allowed to update a given site with multiple process streams (similar to parallel query) as opposed to a single process. Also, the replication functionality has now been built directly into the server software, as opposed to being a series of triggers, for improved performance. Finally, you can now update a snapshot based on the primary key of the table involved, as opposed to using row IDs, which can change in certain databases that require frequent reorganization.

•  Enhanced support for unstructured data. One of Oracle’s initiatives is to extend Oracle to become a media server. To do this, the large object types have been enhanced in size, breaking the old 2GB limitation. Also, you are now allowed to have more than one large object type per table. Finally, the access and update functionality has been enhanced to be more efficient.

•  Optimizer enhancements. The optimizer is that section of the Oracle software that determines which algorithm to use when retrieving information from the database. The optimizer has been improved to make better decisions. One enhancement of special note is the improvement of the algorithms used for a star schema commonyly found in data warehouses.