An exclusive interview with Gordon Mangione, Microsoft vice president of SQL Server

Editor's Note: Gordon Mangione, Microsoft vice president of SQL Server, will deliver a keynote address at the PASS 2001 North America Users' Conference in Orlando, Florida, this month. For more information about the conference, which runs September 18—22, see http://www.sqlpass.org.

At home in an office along the same hallway where he began his career at Microsoft 10 years ago, Gordon Mangione leads a SQL Server team that's at the heart of Microsoft's .NET strategy. Fresh from his role as vice president of the Microsoft Exchange Server team, Mangione is diving into his new role as Microsoft's vice president of SQL Server. In this exclusive interview, Mangione talks to SQL Server Magazine about the success of SQL Server 2000, building a foundation for the future with XML and .NET, and what the next release of SQL Server, code-named Yukon, will bring.

SQL Server Magazine: Where is the SQL Server business today in terms of revenue and market share?

Mangione: This year, SQL Server broke $1 billion in annual revenue. That's a tremendous milestone and real validation not only for SQL Server but also for our customers and the investments that they've made in these systems. We're still seeing fantastic growth in our business. We're well over 30 percent year-over-year growth and taking share from the competition. We're also seeing the growth of databases on the Windows platform outpacing the growth on the UNIX platform 2 to 1. In terms of unit share, SQL Server has always outpaced the competition on the Windows platform; more than 70 percent of all databases deployed on the Windows platform are SQL Server. But this year, we've surpassed competitors both in units sold and in revenue.

SQL Server Magazine: With SQL Server 2000, the price of the database system increased significantly. For example, the price of SQL Server 2000 Enterprise Edition is up about 50 percent compared with SQL Server 7.0 Enterprise Edition. How are your customers reacting to your licensing strategy for SQL Server 2000?

Mangione: Many customers told us they wanted to be able to buy SQL Server on a cost basis that reflected the way they deployed SQL Server in their enterprises. Although we had CALs (Client Access Licenses) and prices for servers combined with those CALs, customers told us they wanted to be able to buy versions of SQL Server that were targeted to the projects they wanted to deploy. As a result, we came out with the per-CPU pricing that lets customers scale their SQL Server pricing along with the cost of the application and the capability of the project that they have. We've found that many customers have embraced the new pricing model. And we're not the only ones who have gone to that model; you're seeing our competition move toward that model as well.

SQL Server Magazine: Do you think that SQL Server 2000 pricing has affected upgrades from SQL Server 7.0 and 6.5?

Mangione: I don't believe we've seen any impact on upgrades from a licensing perspective. If anything, we're seeing the upgrade rate to SQL Server 2000 far outpacing upgrade rates that we've seen for previous SQL Server releases.

SQL Server Magazine: SQL Server 2000 is one of Microsoft's first-generation .NET Enterprise Servers. But a lot of confusion still surrounds Microsoft's .NET strategy and where SQL Server fits in. Can you help clarify what .NET means to the SQL Server community?

Mangione: SQL Server is now XML-enabled and is the foundation server product upon which you build your .NET XML Web Services. SQL Server is at the heart of storage for all XML Web Services, which are at the heart of .NET. With .NET, we're opening up our platform to a whole set of services and devices by embracing XML standards that the industry has defined. This strategy differs from the strategies we've had in the past in that so much of application development with .NET is based on interoperability with other devices and other services in a clear, documented, standard way. Web Services provides a well-defined way to develop, deploy, and manage applications on the Web and to open up these services to other applications. By combining all these Web services, you can build complex applications by using very simple, industry-standard concepts. We've also done work with tools, such as the SOAP \[Simple Object Access Protocol\] Toolkit, to help make it even easier for you to develop XML services.

SQL Server Magazine: With XML at the core of the .NET strategy, will SQL Server evolve in an un-SQL direction? We've heard of some database vendors that are building database-management systems based on XML. But the verbose nature of XML doesn't seem to lend itself to an efficient database system. For example, XPATH queries read XML documents sequentially, and many XML functions don't have the efficient processing of a SQL-specialized database. With all the sophistication in today's relational database management systems (RDBMSs), is the idea of XML as a DBMS a step backward?

Mangione: Hundreds and hundreds of man-years of effort have gone into tuning and building relational database systems to make them the most efficient ways to store, retrieve, and query information. We believe that the right way to use XML in an RDBMS is to actually let you store XML deeper in the database, which enables you to run queries against the database but also to take advantage of all the richness that's within the database itself. So many developers today understand how to build SQL applications, and we're going to continue to make massive investments in that area.

SQL Server Magazine: So, XML isn't a database replacement but rather another data-access technology that works with the database?

Mangione: Absolutely. And I'd go further and say that a lot of people have XML documents that they'd like to store in the database. And we're going to make it easier for them to take advantage of all the technology that's in SQL Server to store those documents, back them up, retrieve them, query them, and index them so that people can better manage their disparate systems.

SQL Server Magazine: The SQL Server team is quickly rolling out enhancements to the database system's XML functionality through Web releases. The first Web release came out in January, and Web Release 2 is in beta now. What's the planned schedule for the Web release program, and what XML enhancements will Microsoft be adding through this program?

Mangione: XML development is moving very fast across the industry. We have a team completely dedicated to building sets of XML services on top of SQL Server; this team is able to release features much faster than we could ever rev the core underlying database engine. The team has a series of releases coming out in beta. A whole set of those services should be available around the time Visual Studio.NET comes out.

SQL Server Magazine: How many customers have downloaded the Web releases, and what kind of feedback are you getting from them?

Mangione: The download rate for the Web releases is phenomenal—in the hundreds of thousands. We're getting a lot of feedback from customers about how they're using the functionality in the Web releases to develop applications. In addition, we're very involved in the World Wide Web Consortium's (W3C's) standards process. As new XML initiatives come out of the W3C, such as those around XML schemas and XPATH queries, you'll see us implement those initiatives as part of our SQL Server XML effort.

SQL Server Magazine: We've heard some people say that they really don't need to learn XML because XML is going to be more of a plumbing feature implemented under the database covers than something you have to explicitly know about and control. Could you comment on that?

Mangione: If you're building solutions using tools and runtimes that natively understand XML, you certainly won't have to dive into the details of understanding how to put tags here or collections there. It's up to us at Microsoft to build great tools to hide a lot of the complexity from users. When you start building heterogeneous systems and you want to interoperate and especially build connectivity solutions to other systems, the XML contract by which you sign with that other system will force you to dive down to another level and really understand the data that's being passed between systems. But we also have products such as BizTalk Server that aim to make it very simple to exchange XML documents with other systems and integrate them into the SQL Server environment.

SQL Server Magazine: In terms of scalability, SQL Server 2000's support for distributed partitioned views was a major milestone in the history of the product and in Microsoft's ability to dominate the Transaction Processing Performance Council's (TPC's) TPC-C benchmark. But some administrative concerns surround the new functionality, which doesn't seem to be catching on with many customers. How are you planning to address these administrative concerns in the next release of the product?

Mangione: We can improve SQL Server's scalability in a number of ways. Certain systems naturally lend themselves to a scale-out design, which distributes data across multiple servers. As you mentioned, we've shown fantastic scalability through scale-out technology in the TPC-C benchmark. Some customers are tapping in to the scalability that distributed partitioned views offer; others are just partitioning their system within their applications. Customers are also telling us that they want to see great scale-up solutions, where they can have scalability within a single box. Working with Unisys and SAP, we've been able to publish 20,000 concurrent users connected on a single system in the SAP benchmark. This result is phenomenal considering that when we released the product, we supported 7500 concurrent users on a standard 8-way machine. In future releases, you'll see us further address the ease of use and scalability of both scale-up and scale-out designs.

SQL Server Magazine: At Microsoft's TechEd 2001 conference, previews of Yukon created quite a buzz. Can you share the overall design objectives of this upcoming release of SQL Server, some features we might expect to see, and when the release will hit the streets?

Mangione: First, we're going to continue to focus on ease of use and eliminating a lot of the mundane aspects of running the system. Making SQL Server easier to manage and building even more scalable, reliable systems are two of our key mandates for Yukon. We're also focused on making business intelligence easier to implement and use across the environment. I like the phrase "business intelligence for the masses." Make it simple: simple to build OLAP cubes, simple to build up your data sources. In addition, the release will bring easier-to-use transformation services. But what's really going to define the release is the integration of .NET and its Common Language Runtime, or CLR, into the database and the implementation of XML deeper into the database. You'll probably see the next release of SQL Server in early 2003.

SQL Server Magazine: You mentioned the upcoming .NET CLR, which will let developers create applications on SQL Server in different languages. But even today, many developers use inefficient, row-by-row procedural code to interact with the database. If these developers can write SQL Server stored procedures in the same way, we'll likely see even more applications that have performance and scalability problems. T-SQL has a set-based efficiency and is very elegant for a database management system. Is Microsoft going to add set-based constructs to any of the new .NET languages? Or are you going to improve T-SQL to let it interact with the CLR? What's the relationship between T-SQL and .NET?

Mangione: Clearly, tens of thousands of applications are written in T-SQL today, and compatibility with those applications and ensuring that those applications work on future releases is paramount to our success with .NET. Absolutely, you'll see T-SQL continuing to run in and be optimized for SQL Server. Your point about set-based programming and how to extend that into languages such as C# and Visual Basic (VB) running in the CLR is a real focus of what we want to do with Yukon. When we integrate the CLR into SQL Server, we need to develop a connected data model that's fast, efficient, and very high-performance. Customers are also telling us they want a symmetric model for developing applications on the middle tier and within the database. They want a common way to debug the programs and to do source control and data-access calls; they want a common programming technique on the back end. So, we're looking at how to leverage our investments in the CLR on the middle tier to develop better applications within the database itself.

SQL Server Magazine: ADO.NET, the data-access technology for the .NET Framework, brings valuable enhancements to ADO but also raises some performance and scalability concerns. One of ADO.NET's core components is the DataSet object, which is basically an in-memory data store with its own set of programs, keys, and referential integrity. These features are valuable in a lot of ways, but programmers who don't have a lot of database experience could use these features to produce applications that don't perform or scale well. Educating customers to retrieve subsets of the data—only the data they need—instead of trying to do row-by-row processing is very important. What's Microsoft doing to protect against some of the scalability issues associated with ADO.NET?

Mangione: Certainly, inexperienced programmers can write inefficient database applications with ADO.NET. The onus is really on us, Microsoft, to get great coding samples out, to produce great MSDN \[Microsoft Developer Network\] articles, and to offer great training tools to show people how to build highly scalable applications with this technology. It's also important that the ADO technology we develop inside Microsoft sits with the SQL Server team so that we can optimize the development of those applications. In addition, we're looking at customer feedback to see how we can improve subsequent releases of ADO.NET.

SQL Server Magazine: What should database administrators and developers do today to get ready for these .NET changes?

Mangione: With the launch of Visual Studio.NET and even with beta 2, which is available today, customers can start learning a lot about what it means to develop applications in C# or VB. They can start working with ADO.NET and the simplification we've brought to data access. ADO.NET is a great way to start moving a lot of your mid-tier applications to the new programming constructs. And that knowledge will naturally lend itself to the next release of SQL Server, which will natively feature the same sorts of facilities.

SQL Server Magazine: Some people in the database world see SQL Server's hallmark ease of use as a dumbing-down of the database system and don't consider SQL Server professionals, in general, as highly skilled technologists, worthy of premium salaries. How does training and best practices—or the lack of them—in the SQL Server community affect Microsoft and the view of it in the database marketplace? And what's Microsoft doing to improve the skills and reputation of SQL Server professionals?

Mangione: Our goal with ease of use is to let our DBAs concentrate more on working with the software developers—to really become the guardians of the data and the people who analyze and tune applications and database access to get great performance out of the system. When we talk about ease of use, we're really talking about automating a lot of the mundane aspects of managing the system, making sure that you can perform all the operations online without having to take the database offline, making sure that the systems adapt to changes in query patterns, and freeing our DBAs to offer much more value to their companies. We'll continue to invest in utilities such as the Index Tuning Wizard to give DBAs the profiling tools they need to deeply analyze their systems. We've seen tremendous payback by working with our research group here at Microsoft and incorporating the work that they've done and the tools they've used directly into SQL Server to benefit our customers. Another great way that DBAs can differentiate themselves and make themselves even more valuable is by learning more than just the relational system. Learn how to use Data Transformation Services (DTS), how to build business intelligence solutions, and how to build analytic solutions that the business leaders inside your company will really appreciate. The best thing we can do at Microsoft to boost the reputation of SQL Server professionals is to provide training tools for more advanced capabilities on SQL Server and to continue to focus on making SQL Server a great enterprise application that people deploy their mission-critical applications on.