DB2 9 for z/OS Goes GA

IBM hopes that DB2 9’s performance, manageability, and security enhancements will make it a must-upgrade release

IBM Corp. last week officially unveiled DB2 9 “Viper” for z/OS, a Big-Iron complement to the next-generation DB2 9 release for Windows, Unix, and Linux it shipped nearly a year ago.

The new DB2 boasts a bevy of enhancements—including IBM’s pureXML technology, which enables DB2 to natively store both relational and XML data, without decomposing or “shredding” the latter into relational bits. Add to this improved data management features, security and disaster recovery enhancements, and a more consistent user interface (designed, Big Blue claims, to help universalize the DB2 management experience across both mainframe and distributed environments) and IBM officials say you have the makings of a must-upgrade release—with an important caveat or two, of course.

Consider PureXML, which has been a source of considerable contention between IBM and its vendor competitors (see http://www.tdwi.org/News/display.aspx?id=8024).

IBM claims that PureXML makes DB2 the only XML-native database on the market—even though both Microsoft Corp. and Oracle Corp. claim to market XML-native databases. According to Bernie Spang, director of data server strategy with IBM, neither vendor natively stores XML data as XML data in the database, however. Instead, they use one of several techniques, such as XML “shredding” (breaking up XML information and turning it into rows in relational tables, as Microsoft did in SQL Server 2000), or binary large objects (BLOB) and character large objects (CLOB)—to decompose (or aggregate) and store XML information in conventional relational tables.

“The problem with that is that if you want to gain any insight, you have to pull the whole CLOB out, and then you and your application code have to decompose it, or you have to have some intermediate layer to parse it and do the query,” Spang argues.

PureXML, on the other hand, stores XML data as XML data, Spang says. “This is the ability to store XML Structured Data in its natural hierarchical state, as opposed to having to transform it and put it into a relational model,” he indicates. “We’re the only ones that support in the same database both structures natively, both the rows and columns relational as well as the hierarchical XML.”

XML is a garrulous language compared to SQL, and XML processing can be onerously intensive. This begs the question: now that DB2 for z/OS boasts native XML storage capabilities, does IBM plan to extend the benefits of the zSeries Integrated Information Processor (zIIP) to XML workloads, too?

Not quite yet, Spang says. For one thing, he notes, XML processing isn’t typically an issue in the data warehousing and decision-support use-cases Big Blue outlined when it introduced zIIP last year. “XML-based data applied in a warehousing or a deep data-mining application is not the first order of business we’re focused on at this point,” Spang confirms. “When we talk about the type of queries and the deep data mining across the relational data in the database, that’s what we’re first talking about with the zIIP. We’re certainly on a point as to where we’re headed where [XML processing] could be a part of [zIIP-approved workloads]. It’s possible we’ll see uptake and use of XML data in this way even faster than we have been thinking.”

There is a zIIP-friendly tip to the DB2 9 for z/OS release, Spang notes. “We do add support for SQL Procedure Language execution on the zIIP, so there is that additional workload that is applicable for running on the zIIP, and like the others [zIIP-approved workloads] that is part of reducing the cost of operation.”

Elsewhere, Spang continues, DB2 9 for z/OS boasts additional performance and availability improvements, such as improved INSERT performance, logging optimizations, and enhanced sparse index and in-memory caching techniques.

“There are additional optimizations in DB2 9 that are focused on increasing the availability and the speed,” he says. “Another important element is under the heading of governance. Governance is about both security—and DB2 9 offers additional access control, finer-grained access control for that—but it’s also about having the capability to ensure proper accountability, which involves not just the security and the access control, but also the ability to demonstrate … with confidence that you’re properly controlling your data.”

In this last regard, Spang cites DB2 9’s support for role-based access control and “trusted security context” connections. The latter feature lets DBAs specify trusted connections (e.g., Resource Recovery Services, DataSet Name) instead of managing passwords for (and granting users access to) each of those resources.

The revamped DB2 also boasts a more consistent management experience, such that DBAs who are familiar with the distributed flavor of DB2 should feel relatively at home if called upon to manage DB2 for z/OS—and vice-versa, Spang asserts. “[Customers] don’t want to spend more on hardware, software, and staff than I need to achieve [their] goals, so we focus on improving DB2 productivity—improving DBA efficiency—by building more autonomic capabilities into the data server itself,” he comments.

“One of the key points of efficiency is the fact that there’s less requirement for System z-specific skills in [managing] the DB2 data server on system z. There’s greater consistency across all relational data servers, so the skill base that you have to draw from and the amount of system-z-specific knowledge you have to have is far less.”

The DB2 for z/OS and DB2 for distributed systems development efforts are increasingly converging, Spang indicates. “That is an important priority for us. DB2 9 is a significant step forward for us in that regard, and you’ll continue to see that,” he says. “The delivery cycles are somewhat different, but you should continue to see capabilities that were first on the mainframe showing up on the Linux, Windows, and Unix version, and vice-versa.”

Development on the version of DB2 which ships with System i, on the other hand, is still a largely separate enterprise. “We are connected with the System i team, and … we are working on continually bringing the commonality of the experience together. That’s a slightly different delivery process, because [DB2 is] actually integrated within the i5 OS. So that development and that delivery schedule is tied with the priorities of the entire OS. [The System i] platform and its role as an integrated application platform, give it a different set of priorities.”

About the Author

Stephen Swoyer is a Nashville, TN-based freelance journalist who writes about technology.