May 31, 2003

Center of the Universe: The Oracle View

The vision of 21st century data management as told through Web-exclusive interviews with key strategists at IBM, Microsoft, and Oracle

by Ken North

Andrew Mendelsohn, senior vice president of database server technologies at Oracle, shared his perspectives about key data management issues with Ken North. The discussion covered Oracle's view of SQL and XML integration, the importance of integrating the database with message queuing and business analytics, the future of grid data access, and the value of multiple platform support.

North: XML and metadata technologies make it possible to query distributed data sources and aggregate or analyze data using middleware. Is it better to support business analytics with middleware or to move the analysis into the database?

Mendelsohn: Whenever possible, we advocate moving the analytic processing as close as possible to the data. In Oracle9i Database, we have powerful SQL analytics, OLAP analysis, and data mining algorithms integrated inside the database engine. This has many benefits. There's improved performance by leveraging the scalable database query-processing engine and minimizing data movement between separate engines. There's reduced complexity by having one system to install, upgrade, and manage. There's vastly improved data security by minimizing data movement and the number of separate systems managing the data.

When there's a need to integrate data across multiple Oracle or non-Oracle databases, both the Oracle9i Database and middle-tier technologies such as the Oracle9i Application Server may be used. Oracle9i Database can federate multiple Oracle and non-Oracle databases. It can mask the physical location of the data and automatically manage distributed transactions. It also provides enhanced optimization features to efficiently access and manipulate data in both local and remote databases. Alternatively, applications running in the middle-tier may use technologies such as ODBC, JDBC, and OCI [Oracle Call Interface] to directly access and integrate data from multiple data sources.

North: In the 1980s, Oracle built a reputation as the DBMS that was available for the largest number of operating systems. Going forward, how important is it for Oracle to support divergent platforms such as Linux, J2EE, Windows, and .Net?

Mendelsohn: Portability is as critically important to our customers today as in the past. They want to write their applications once and then transparently deploy them across SMPs, clusters, and the blade server platforms just coming to market. Oracle9i Database with Real Application Clusters is the only database technology available that allows customers to do this for all their custom and packaged applications. Customers also want to be able to efficiently use both J2EE and .Net application development platforms.

We're seeing huge interest from our customers and ISVs [independent software vendors] in Linux. As always, we use the same Oracle code base across all operating system platforms. This ensures customer database applications can move unchanged to Linux. Also, Oracle is providing a single point of support for our Linux customers using Red Hat Advanced Server or UnitedLinux. Oracle fixes all bugs in Oracle products and priority 1 bugs in Linux; lower priority Linux bugs are passed on to Red Hat and UnitedLinux for resolution. Oracle is also making a significant investment in the Linux open source community. We have contributed the Oracle Cluster File System to the Open Source Initiative and have also worked closely with the Linux community to enhance the operating system kernel to be a more robust database platform.

North: There's an ongoing debate about the impedance mismatch between objects, object-oriented languages, and SQL databases because of different type systems, different computational models, and so on. Do XML, XQuery and XPath present an impedance mismatch problem for SQL? Do we need an XML-centric database programming language?

Mendelsohn: Oracle's approach to this impedance mismatch has been to converge the SQL and XML technologies. Starting in 1997 with the release of Oracle 8.0, we added object constructs to SQL in what was called object-relational technology. These constructs were standardized in the SQL:1999 standard. The expressive power of an object-relational SQL schema is quite comparable to XML Schema. This gave us the ability to rapidly add native support for XML into the Oracle9i Database. XML Schema and XMLType are native to Oracle9i Database, as are XSLT transformations and XPath traversal. There's no information loss in storing XML, so there's no impedance mismatch. No other database has native support for XML Schema and XMLType. The result is excellent performance for Oracle9i Database.

You can program with Oracle using an XML-centric view, and you can program with SQL as you always could. Oracle9i Database enables XML-SQL duality — the ability to perform XML operations on relational data and relational operations on XML data. This gives you the best of both worlds. For example, you can seamlessly perform SQL-style OLAP on XML documents, and you can also perform XPath queries over relational data. We have extended the relational system in a way that makes XML fit in neatly. No other vendor can match the comprehensiveness of our approach.

North: Oracle supports XML messaging and different forms of distributed computing, including Web services and grid services. Do these technologies present querying and load balancing problems?

Mendelsohn: While Oracle supports XML and different forms of distributed computing, we provide unique technology that eliminates querying and load balancing problems. There's no need to partition data in a grid environment. Oracle9i Real Application Clusters can efficiently distribute a database workload across multiple servers without partitioning the data. This is a critical point. If you must partition the data, you can't easily grow or shrink the number of servers operating on your data. If you can't dynamically provision resources to your database, you won't realize the improved resource utilization benefits that grids can offer.

Oracle also provides key features to support partitioning data across multiple servers, should you choose. Our federated database technology masks the physical location of data; this makes it simple to update or query data in a distributed environment. Oracle's powerful query optimizer automatically determines the best execution and access plan for any distributed query or update. Our transportable tablespace features allow customers to effortlessly move or copy a subset of a database to another server for processing. We can then maintain that copy, keeping it in sync with the original database using Oracle Streams. And while we support Web services for customers to build tightly coupled distributed applications, Oracle Streams database-integrated message queues enable building loosely coupled distributed applications, which dramatically improves the robustness and resiliency of these applications.

North: Asynchronous messaging and queuing are important building blocks for developing scalable, distributed applications. How important is it to have messaging software and queues tightly bound with databases?

Mendelsohn: Integrating messaging software and queues into the database provides many development and operational benefits. Most business messages originate from databases, and eventually update databases. Integrating the database and message queuing software means the developers only need to learn a single product. They only need to manage a single security and transaction model. The data types used for the database are the same as those supported by the message queues, again simplifying development. Integration also brings many operational benefits. There's only a single product to install and manage. Because queues are in the database, they inherit all the security, high availability, and recoverability features of the database, ensuring that no data will be lost — even if it's in transit — in the event of a failure or security breach. Finally, there are performance benefits. Middleware message queuing solutions must use an expensive distributed two-phase commit to transactionally store data in the destination database. Because Oracle's message queues are already in the database, there's no need to perform this expensive operation.

• Privacy
• Your California Privacy Rights
• Copyright © 2008 United Business Media LLC