The database systems vendors profiled in this appendix have been selected because of their unique positions within the broader database industry. They include the providers of the leading enterprise-class DBMS products, some smaller companies that are leaders in new technology areas, pioneers in newer segments of the database market, and vendors that focus on embeddable database technology. Any compilation like this cannot possibly be exhaustive, and the omission of a company does not mean that its products or capabilities are inferior to those of the vendors profiled here. Collectively, these companies and their profiles as presented, illustrate the landscape of today’s multibillion-dollar database software and services market. The vendors are:
- A2i, Inc.
- Arbor Software (now Hyperion Solutions Corporation)
- Birdstep Technology
- Computer Associates (Jasmine, Ingres)
- Computer Corporation of America (Model 204)
- Empress Software
- eXcelon (ObjectStore, XIS)
- Gupta Technologies (SQLBase)
- Hewlett-Packard (NonStop SQL)
- IBM Corporation (DB2, Informix, Cloudscape)
- Informix Software (now part of IBM)
- Microsoft Corporation (SQL Server)
- MySQL AB
- Objectivity
- Oracle Corporation (Oracle, Rdb/VMS)
- Persistence Software
- Pervasive Software
- PointBase
- PostgreSQL
- Quadbase Systems
- Red Brick Systems (now part of Informix Software)
- Sybase, Inc.
- TimesTen Performance Software
- Versant Corporation
1. A2i, Inc. (www.a2i.com)
Founded in 1993, A2i develops and markets an integrated, database-driven, cross-media catalog publishing system that centralizes the management of catalog data, simplifies the catalog production process, and completely automates the catalog production workflow. The system includes tools for creating, designing, and publishing both printed and electronic catalogs; supports single source cross-media publishing to paper, CD-ROM, and the Web; and efficiently manages catalogs containing from hundreds to millions of items.
All of A2i’s software products layer on top of a SQL-based DBMS. They include performance accelerators that improve catalog access by a factor of 10 to 1000 times that of SQL alone, and feature additional catalog-specific functionality that supports interactive browsing and sorting of large databases in ways that would otherwise be impossible using a traditional SQL-based DBMS alone. A2i’s parametric search technology—an alternative to DBMS-style query forms that is intuitive, easy to use, and very, very fast—allows a user to search an entire catalog and locate any item or group of items in a matter of seconds, narrowing down from thousands or millions of items to one or several with just a few mouse clicks.
2. Arbor Software (www.hyperion.com)
Arbor was one of the early leaders in the development of Online Analytic Processing (OLAP) databases and tools. Arbor’s flagship Essbase OLAP server was first introduced in 1992 and pioneered many of the capabilities that have now become commonplace in analytic systems. Large corporations typically use the Essbase product suite to create integrated reporting, analysis, and planning systems.
Current versions of the Essbase product support both client/server and web-based analytic processing and reporting. They also support both precalculated data (a hallmark of most OLAP systems) and dynamic, on-the-fly calculations. Another major enhancement of the Essbase product is distributed OLAP capability, which allows OLAP databases to be partitioned across computer networks. Essbase supports both its own proprietary multidimensional database formats and integrates with conventional relational databases. The latest version, marketed under the brand name Essbase XTD, is positioned as a business intelligence platform, and runs across a number of different operating systems.
In 1998, Arbor merged with Hyperion Solutions Corporation to create a $500 million company (annual revenue) focused on business reporting and analysis. The product line has grown to include a substantial business in integration products and customization services. It spans applications from single-user analysis on Windows workstations to enterprisewide web-based OLAP deployments for hundreds of users.
3. Birdstep Technology (www.birdstep.com)
Birdstep Technology is a Norwegian software company focused on wirelessly connected systems and embedded solutions. Birdstep entered the database business with its September 2001 acquisition of Raima Corporation. Raima, founded in 1982, was an early database vendor focused on the IBM PC database market. Its initial db_VISTA product was first released in 1984. It has been steadily enhanced over the years and combined with an object manager to create the Raima Database Manager (RDM) product, now marketed by Birdstep.
A newer Raima product, the Velocis Database Server, was first shipped in 1993. Velocis is a SQL-based relational database system with an ODBC interface. It is designed as an embeddable database, and the company targets it to professional application developers—independent software vendors (ISVs) and value-added resellers (VARs) who use it as a bundled database foundation. Velocis runs on Windows, Windows NT, OS/2, and many UNIX-based operating system variants.
A distinctive feature of the Velocis server is its explicit support for network data model’s embedded pointers within a SQL-based database. A CREATE JOIN statement specifies an explicit relationship, implemented with network database-style pointers, which are stored within the database structure. These can then be exploited with SQL syntax, delivering very fast performance. Velocis supports C/C++, Java, Visual Basic, Delphi, and Perl language interfaces as well as the industry-standard ODBC interface.
Birdstep sells the Velocis product as a server edition of the Raima Data Engine.
A complementary Mobile Edition is tuned for use in mobile devices with wireless connectivity. A third edition is focused on embedded applications.
4. Computer Associates (www.cai.com)
Computer Associates (CA) is one of the world’s largest independent software companies. Initially focused on mainframe software, the company has steadily expanded its focus to provide an extensive line of software products and services for enterprise data processing. Computer Associates has been built largely through acquisition, taking advantage of its large direct sales force and well-established relationships with senior Fortune 500 information systems executives. Through its acquisitions, it has steadily added more products to its portfolio.
Ingres, one of the earliest relational database systems to appear on the market, is now a product of Computer Associates. It was originally developed at the University of California at Berkeley as a research project under the direction of Professor Michael Stonebreaker. The research project became the foundation of an independent company, which eventually changed its name to Ingres Corporation in the 1980s. Ingres and its native QUEL query language were an early competitor to SQL, which was backed by rival Oracle Corporation. Although most analysts gave Ingres clear claim to technical leadership, Oracle’s aggressive marketing and sales efforts, coupled with IBM’s backing of SQL, eventually led to SQL dominance in the market. Eventually, Ingres was adapted to support SQL, which emerged as the dominant standard. In the 1990s, Ingres was sold to the ASK Group, and eventually to Computer Associates.
The current version of the product, Advantage Ingres Enterprise Relational Database, is a comprehensive relational database management product suite. The core Ingres/DBMS is augmented by a capability that links the DBMS to the Web. The product includes standards-based ODBC access, a sophisticated distributed data manager, and built-in XML support. It runs on major UNIX server platforms and Windows server versions.
Computer Associates also offers Jasmine Object Database, a new object-oriented DBMS. Although touted as a complete DBMS solution with a modern object-oriented architecture, two major areas of focus for Jasmine are multimedia and Internet applications. The core DBMS is heavily object-oriented, featuring multiple inheritance, instance and class methods and properties, and set-level methods. Methods for the Jasmine object-oriented DBMS can be written in C, C++, or Java. Jasmine includes an extensive class library with support for multimedia data types (images, animation sequences, audio, video, rich text, page layouts). CA is clearly positioning Jasmine as a new- generation, pure object-oriented database. It is not positioned as having object/relational capabilities, and it does not offer any SQL access to its own data management capabilities. CA does tout Jasmine’s integration with back-end relational databases (Oracle, Sybase, Informix, SQL Server, DB2) and mainframe files (VSAM and CA-IDMS). The linkage to an Ingres back-end is especially close, with tightly integrated transaction management, security, and replication management capabilities.
5. Computer Corporation of America (www.cca-int.com)
Computer Corporation of America (CCA) is one of the pioneering software companies, and has been involved in data management since its founding in 1965. It develops and sells one of the earliest DBMS systems: Model 204. The product has been substantially enhanced over the years, but the focus continues to be on mainframe systems.
Model 204 features an ANSI-compliant SQL interface, even though the underlying structure is a network database architecture. The network structure is manifested in Model 204’s embedded table capability—essentially a table-within-a-table structure. Although network databases fell out of favor with the advent of SQL and the relational model, some of the same capabilities provided by the network systems are now appearing in highly touted new object-relational systems. The nested table structure offered by Model 204 is an example of such a capability, which appears in object-relational systems from Informix and in Oracle’s flagship Oracle8 object-oriented extensions.
The current version of Model 204 includes multiprocessing and parallel query options for data warehousing applications. Over the years, its indexing structures have become quite sophisticated and now include bitmap, hashing, B-tree, and record list schemes. Another unique feature of Model 204 is support for iterative queries—queries that are carried out against the results of previous queries. SQL-based access to mainframe Model 204 databases is available through CCA’s Connect* product, which
offers ODBC and OLE-DB APIs for remote database access from Windows and UNIX-based client workstations.
CCA also provides System 1032, a high-performance database product for the OpenVMS operating system (the successor to Digital Equipment’s popular VAX/VMS from the 1980s and 1990s). System 1032 focuses on high-performance query applications. Although not originally a relational database, it now offers access via ODBC and SQL.
6. Empress Software (www.empress.com)
Empress Software produces an ANSI SQL relational database system for embedded applications. The company was founded in 1979 in Toronto, Canada, and currently is headquartered outside Washington, DC. The Empress DBMS offers both an ODBC callable API and Embedded SQL interfaces. It also offers a low-level set of database access calls that come in below the SQL access layer. These calls provide direct access to the Empress storage manager layer for very high performance record insert, update, delete, and retrieve operations.
The Empress DBMS runs on many different UNIX-based systems, including several UNIX operating system variants that run on Intel processor-based systems. It also supports Windows, Windows NT, and a range of real-time operating systems typically used for embedded applications. It offers a rich collection of data types, plus user- definable functions and procedures. For Internet-based applications, Empress also offers script language interfaces for the popular Perl and Tcl/Tk scripting languages.
7. eXcelon (www.exln.com)
eXcelon was founded as Object Design, one of the early object database vendors in 1988 in Burlington, Massachusetts. The initial version of its ObjectStore object database system was shipped in 1990. The ObjectStore database can be used either as a standalone object-oriented database (OODB) or as a caching solution, providing fast, object-oriented access to data that has been pulled forward from a back-end relational database such as Oracle or DB2.
In the late 1990s, as the market momentum of the object-oriented databases stalled, Object Design invested in an XML database product, named eXcelon, and eventually renamed itself after this product in February 2000. After roughly a year, the company moved to reassure its installed base of ObjectStore customers by re-creating an Object Design division, responsible for the ObjectStore product line, which remains a major product of the company today. In May 2001, eXcelon again moved to broaden its offerings through a merger with C-bridge, a professional services company. Today, its products include the original ObjectStore OODB; an XML database named XIS; a caching product focused on application server data caching named Javlin; and professional services for solutions development and deployment based on these technologies.
8. Gupta Technologies (www.guptaworldwide.com)
Gupta Technologies was founded by a former manager of Oracle’s microcomputer division, Umang Gupta. The company’s initial focus was a DBMS and database development tools for PCs and PC local area networks. Renamed Centura Software in the late 1990s, the company has returned to its original name, and now focuses on embedded database software and development tools, primarily targeting independent software vendors and value-added resellers.
SQLBase, the company’s flagship DBMS product, has evolved considerably since its origins as a stand-alone and client/server database for IBM PCs under MS-DOS. It has grown to support Windows NT and NetWare as database servers. Centura currently targets SQLBase for applications on PCs and sub-PC devices such as handheld PCs, RISC-based information appliances (e.g., smart phones), and even smart cards. It features a small footprint, zero-maintenance operation, and a scalable architecture. ODBC 3.0 and JDBC interfaces are provided. Companion products complement the core SQLBase offering with software development tools, report writing, mainframe database connectivity, and similar capabilities.
9. Hewlett Packard (www.hp.com)
Hewlett-Packard’s NonStop SQL database products are a result of its merger with Compaq in 2002. Several years earlier, in 1997, Compaq (at the time a vendor of Intel-architecture PCs and servers) acquired Tandem Computers, an early leader in the market for fault-tolerant minicomputer systems. Many Tandem systems are used by financial services and transportation companies for use in online transaction processing applications that demand 24/7 nonstop operation. For example, Tandem systems run many of the major banks’ ATM networks and many of the leading stock exchanges. Tandem’s older systems ran the proprietary TXP operating system, and fault-tolerant applications are generally written in the proprietary Tandem Application Language (TAL). More recent Tandem systems are based on UNIX operating systems.
Database management for nonstop applications on Tandem systems has been provided for many years by a SQL-based Tandem-developed RDBMS, called NonStop SQL. Because of Tandem’s heavy OLTP emphasis, NonStop SQL has pioneered several special techniques, such as disk mirroring. It also takes advantage of the inherent Tandem multiprocessor architecture and provides distributed database capabilities.
The programmatic interface to NonStop SQL is through embedded SQL.
During the 1980s and early 1990s, virtually every minicomputer vendor had its own proprietary SQL-based implementation (Digital with Rdb/VMS, Hewlett-Packard with Allbase/SQL, Data General with DG-SQL, and so on). Over the years, all of the other systems vendors have concluded that the high cost of maintaining their own RDBMS with
competitive features was prohibitive. They also had difficulty managing the dual roles of competing with the independent DBMS vendors (such as Oracle) and also working with them as ISV partners on their platforms. As a result, Hewlett-Packard (via its acquisition of the former Tandem product line) is the only remaining major system vendor (except for IBM) with its own proprietary SQL-based RDBMS.
NonStop SQL is still an important product for the Tandem customer base. It ships in two versions. NonStop SQL MP is a distributed, highly parallel database system, designed to run on systems with 2 to over 4000 processors. NonStop SQL MX is a version of NonStop SQL extended with object-relational capabilities.
10. IBM Corporation (www.ibm.com)
IBM, the largest computer company in the world, is also among the largest software vendors in the world. IBM researchers pioneered the relational database concept, invented the SQL language, and produced the first relational database prototype— System/R—in the 1970s. Over the next two decades, IBM’s flagship relational database—DB2—for its mainframe systems, pioneered several relational capabilities that have since found their way into mainstream RDBMS products and into generations of SQL standards. During this same time, relational database technology proliferated onto other IBM computer system platforms, including time-sharing mainframes, minicomputers, UNIX-based workstations and servers, and personal computers. In the late 1990s, IBM moved aggressively to bring all of these IBM data management products under a single umbrella (using the DB2-Universal Database name), and to offer its DB2 relational database technology on non-IBM platforms from other leading UNIX system vendors.
In addition to its own database development, IBM expanded its database reach substantially with its 2001 acquisition of the database-related businesses of Informix Software. Informix had been a UNIX-based DBMS pioneer, with its internally developed flagship Informix product. Informix had also been an aggressive acquirer of other database companies and technologies, including object-relational database pioneer Illustra (1996), data warehousing pioneer Red Brick (1998), database vendor Ardent Software (1999), and Java database pioneer Cloudscape (1999). Many of these database products survive as part of IBM’s data management software product line.
Today, DB2-Universal Database is a comprehensive, enterprise-class, SQL-based relational database system. DB2 implementations run on a very broad range of platforms, from desktop personal computers to the largest IBM mainframe clusters. DB2 can be characterized as a quite complete and comprehensive SQL implementation, especially in areas that have been traditional IBM strongholds, such as high availability, reliability, maintainability, and worldwide support (international character set). Adjunct products and tools support software development, distributed database capabilities, data warehousing, data replication and distribution, and most other major areas of database activity. Although IBM has made its products available on non-IBM platforms, the vast majority of IBM DB2 installations are on IBM computer systems and are sold as part of integrated IBM-based enterprise systems, and the core strength of DB2 is on IBM mainframe systems.
The pioneering object-relational DBMS, Illustra, survives in IBM’s product line as the Informix Dynamic Server, and retains a significant market share on UNIX-based systems. The Cloudscape product is available from IBM, distinguished as a 100 percent Java database especially tuned for mobile computing applications. Red Brick data warehouse provides a specialized business intelligence server, but its functions overlap considerably with BI and OLAP tools in the DB2 product line. UniVerse is focused on client/server and web-based data management, with low management overhead, but also has considerable overlap with other IBM products. Finally, it’s worth noting that IMS (a non-SQL, hierarchical database whose origins predate the relational model) remains a very significant IBM data management product, with ongoing development.
11. Informix Software (See IBM Corporation)
Informix was one of the original leaders in the UNIX-based relational database market. The company’s first relational DBMS was implemented on UNIX-based microcomputer systems in the early 1980s, and was known for its efficiency and compactness. In 1985, Informix was rewritten as a SQL-based DBMS and introduced as Informix-SQL. It was subsequently ported to a wide range of systems, from IBM PCs under MS-DOS to Amdahl mainframes running UNIX. Informix was also one of the first database vendors to expand its product offerings beyond the core database engine to include development tools. Its Informix-4GL product family supports the development of forms-based interactive applications.
In the early 1990s, Informix expanded its product line into the office automation area, including among other products, a database-integrated spreadsheet named Wingz. This effort was not very successful against Microsoft’s office suite juggernaut, and Informix refocused on its core database capabilities. One of its flagship products during the mid-1990s was Informix Parallel Server, the technology leader in so-called parallel query technology. Parallel Server splits the processing of a single complex query into multiple, parallel operations, which can take advantage of symmetric multiprocessing (SMP) servers. Later, Informix established a leadership position in object-relational technology through the acquisition of Illustra. Illustra was a venture- backed database software firm, led by Michael Stonebreaker (the same Berkeley professor who had led the development of Ingres years before). A side-effect of the Illustra acquisition was a proliferation of product lines and development teams within Informix, adding to some confusion among Informix customers.
As the enterprise database market turned into a three-horse race in the late 1990s, Informix found itself a much smaller competitor than the “Big 3” (Oracle, IBM, and Microsoft). Informix resources were also split between its original data management business and an emerging and faster-growing business in enterprise data integration
tools. Informix sold the database part of its business to IBM in April 2001, for more than $1 billion. The data integration part of the business was renamed Ascential Software, and continues with a focus in that market.
12. Microsoft Corporation (www.microsoft.com)
Microsoft Corporation, the world’s largest personal computer software company, is also a major vendor in the SQL-based database market. Microsoft’s first foray into database products came in 1987 and began as a defensive move. With the announcement of OS/2 Extended Edition, IBM tried to establish built-in database management and data communications as key components of an enterprise-class PC operating system. In 1988, Microsoft responded with SQL Server, a version of the Sybase DBMS ported to OS/2. Although Microsoft later abandoned OS/2 in favor of its own Windows NT operating system, SQL Server continued as its flagship DBMS. Today, SQL Server is a major product in the workgroup database segment, and Microsoft is aggressively moving to establish it as an enterprise-class DBMS competing with Oracle and DB2.
Expanding on its early experience with SQL Server, Microsoft moved on several other fronts to expand its role as a database vendor. In the early 1990s, Microsoft acquired Foxbase Corporation, developer of the Foxbase DBMS. Foxbase had established itself as a very successful clone of dBASE, the most popular and widely used PC database product. Through the acquisition, Microsoft moved to challenge Borland International, which had acquired the rights to dBASE shortly before.
While the Foxbase acquisition was focused more on the PC-installed base and the relatively mature market for character-based, flat file PC databases, Microsoft’s internal development focused on the new, growing market for graphical lightweight relational PC databases. After several false starts and abandoned development prototypes, the result product, Microsoft Access, was introduced. Microsoft Access continues today as both a stand-alone lightweight database product and a front-end for SQL-based production databases.
Microsoft also moved aggressively to enable Windows as a database access and database development platform. Its first major move in this area was the introduction of Open Database Connectivity (ODBC), a SQL-based API for database access. Microsoft built ODBC capability into Windows and successfully lobbied the SQL Access Group, a database vendor association, to adopt it as a callable database API standard. This early version of ODBC eventually made its way into the formal ISO standards as the SQL Call-Level Interface (CLI). Microsoft has continued to evolve ODBC and expand its capabilities.
Microsoft has also layered other database access APIs on top of ODBC. The first such step was to incorporate database access into Microsoft’s Object Linking and Embedding (OLE) framework for linking applications together. The OLE/DB portion of the OLE suite provided source-independent data access, and relied on ODBC as its underlying architecture for working with relational databases. Later, with the recasting of OLE into the ActiveX component framework, another layer was added to the database access hierarchy. The Active Data Objects (ADO) set of components provide data access within Microsoft’s Component Object Model (COM) architecture. Again, the ADO capabilities are layered on top of ODBC for relational database access.
Paralleling the evolution of the Windows database access capability, Microsoft has steadily expanded and enhanced the capabilities of SQL Server. SQL Server 7, introduced in 1998, represented a major step forward. SQL Server 2000 continued this trend, and came with a parallel marketing push to establish it as an enterprise-class database product. SQL Server features have grown to include an integrated OLAP server and business intelligence capabilities, putting Microsoft squarely into competition with the data warehousing vendors and the warehouse-oriented database capabilities of Oracle and DB2. The high-end Enterprise Edition package provides failover clustering, multiprocessing support for up to 32-way SMP systems, and much more extensive replication services for both online and offline distributed databases, and more recently, XML support as part of Microsoft’s .NET web services initiative. The debate continues over SQL Server’s readiness for enterprise-class data management, but Microsoft continues to work, release by release, toward that goal.
13. MySQL AB (www.mysql.com)
MySQL AB is the Swedish company that publishes and distributes the most popular open source SQL database, MySQL. Founded by two Swedes and a Finn (David Axmark, Allan Larsson, and Monty Widenius), the company’s mission is to “make superior data management affordable to all.” The company is owned primarily by its founding partners, and is firmly devoted to the open source model. The MySQL software, including source code, is free for noncommercial use under the GNU license.
The partners have grown MySQL into a virtual company, with employees distributed around the world. The company earns its revenues from support, training, and other services associated with MySQL, and from revenue-bearing commercial licenses. Several million free copies have been downloaded. The MySQL product is distinguished by being fairly compact in size, and delivering high performance. The capabilities of the product continue to grow, with an active user base that contributes enhancements. Monty Widenius, the principal author of the original MySQL product, acts as moderator for the contributions and orchestrates the formal releases from MySQL AB.
14. Objectivity (www.objectivity.com)
Objectivity was one of the early object-oriented database vendors, and has steadily enhanced its Objectivity/DB OODBMS over the years. It has added fault-tolerant and data-replication capabilities to its core object database engine. Access to Objectivity/DB is provided from C++, Java, and SmallTalk.
Although Objectivity remains firmly focused on an object-oriented architecture, it has moved to provide SQL-based access to its object database engine, through both ODBC and proprietary APIs. The SQL language used through these interfaces contains many extensions to accommodate access to object database structures. Unique object-ids within the Objectivity/DB database are automatically mapped to row-ids available via the SQL interface. Object associations within the OODB are available for use as SQL join criteria. Stored procedures and triggers are presented via extended SQL features. Extended SQL syntax is also provided to access elements of arrays and nested object structures, which appear as complex columns to the SQL user. These capabilities provide the advantages of SQL-based access to many of Objectivity/DB’s object-oriented capabilities, but at the expense of very nonstandard SQL syntax.
The Objectivity/DB product is focused on managing complex data across a broad range of operating environments. It is deployed in large enterprise databases providing access for hundreds of users to terabytes of data. At the other extreme, Objectivity/DB is positioned as an embeddable component for data management within real-time systems.
15. Oracle Corporation (www.oracle.com)
Oracle Corporation was the first DBMS vendor to offer a commercial SQL product, preceding IBM’s own announcement by almost two years. During the 1980s, Oracle grew to become the largest independent DBMS vendor. Today, it is the dominant enterprise DBMS competitor, selling its products through an aggressive direct sales force and through a variety of other channels.
The Oracle DBMS was originally implemented on Digital minicomputers, but the center of gravity of Oracle system sales shifted firmly to UNIX-based minicomputers and servers in the 1990s. One of the major advantages of Oracle is its portability. It is available on dozens of different computer systems, from Windows-based laptop computers through Sun, HP, and IBM UNIX-based systems to IBM mainframes. Using Oracle’s networking software, many of these Oracle implementations can participate in a distributed network of Oracle systems. With these capabilities, Oracle has targeted enterprisewide database deployments and has been effective in leveraging its market leadership into a position as an IS-imposed corporatewide database standard in many organizations.
The Oracle DBMS was originally based on IBM’s System/R prototype, and has remained generally compatible with IBM’s SQL-based products. In recent years, Oracle has been aggressively marketing the OLTP performance of its DBMS, using benchmark results from multiprocessor systems to substantiate its claim as the OLTP performance leader. One round of advertisements in computer industry publications touted a breakthrough level of 100,000 TPC-C transactions per minute on a high-end cluster of SMP 64-bit Digital Alpha servers.
Oracle has consistently combined good technology with an aggressive sales force and high-profile marketing campaigns (including the high-profile presence of its flamboyant CEO and founder, Larry Ellison). It has expanded its product line to include not only DBMS software and database development and management tools, but also enterprise applications software for financial and business management applications. Oracle’s core server products also include an application server for implementing multitier Internet applications.
Oracle also acquired the Rdb relational database from Digital Equipment Corporation, picking up a large installed base of Digital users that it is converting to its Oracle products. Consulting services and recurring maintenance revenues have also become a major part of its revenue. It has also announced that it will make several of its products available on an outsourced basis, effectively allowing customers to use them on a fee- for-services basis. Today, DBMS licensing revenues account for less than half of Oracle’s annual revenues, but enterprise-class data management remains at the heart of the company’s business.
Oracle8 and Oracle8i, introduced in 1998 and 1999, respectively, and Oracle 9i, introduced in late 2000, represented major steps forward in the evolution of the Oracle DBMS. They feature extensive object-relational capabilities, including abstract data types, object structures (such as nested tables, arrays, and sequences), Java APIs (both embedded SQL for Java and a JDBC callable API), and specialized capabilities for high-performance OLTP on SMP systems and data warehousing. To accommodate a broad range of systems, low-end DBMS capability continues to be provided by an Oracle Light product for notebook systems. Oracle 9i is specifically focused on integration of the Oracle DBMS with Internet technologies, such as web and application servers. In addition, marketing for Oracle9i has placed a strong focus on reliability and scalability, with claims of being unbreakable.
Oracle considers its major competitor to be Microsoft, and it embraces a networkcentric enterprise computing architecture to combat Microsoft’s PC-centric view. In the Oracle view, a centralized database system is the critical data store for all information within an organization, which should be accessible anytime and anywhere via the Internet. Easier central control and administration provided by this architecture are key selling points for Oracle to enterprise IS organizations.
Oracle is also aggressively pursuing a one-vendor approach to enterprise IT shops. With the introduction of the Oracle applications suites in the 1990s, Oracle became a competitor to enterprise application vendors like SAP, BAAN, and PeopleSoft. The addition of the Oracle application server put Oracle into competition with BEA and Sun’s application server, as well as IBM’s WebSphere. With its suite of application, database, and middleware products, Oracle’s message to corporate IT organizations is that the best way to deliver applications to their company is with an all-Oracle approach.
16. Persistence Software (www.persistence.com)
Persistence Software was initially focused on software that bridged the gap between object-oriented development and messaging technologies (including object request brokers) and relational database technology. Its middleware products supported object-based data management structures and requests, and mapped them into relational databases stored in the major RDBMS systems. One of the primary target markets for Persistence products has been the financial services market.
Over time, Persistence enhanced its products and repositioned them as a transactional application server. The company’s PowerTier server family includes versions designed to support C++ development or Java (via Enterprise JavaBeans). One of the major features of the PowerTier servers is in-memory caching of objects. Other capabilities of the servers include object transaction isolation and object triggers. The servers continue to offer database independence, integrating with the mainstream enterprise database engines of Oracle, Informix, Sybase, and Microsoft. Application development in C++, Java, and Visual Basic is supported.
More recently, Persistence has repurposed its caching technology and packaged it into two different products. Persistence Dynamai is a dynamic caching product designed to speed web browsing by caching dynamically generated web pages. Persistence EdgeXtend is a data cache for application servers, designed to speed their operation in database-intensive applications. It works with BEA WebLogic and IBM WebSphere.
17. Pervasive Software (www.pervasive.com)
Pervasive Software traces its roots back to the earliest days of personal computer databases. The storage manager that underlies the Pervasive products, Btrieve, was initially developed as a PC-based database for MS-DOS systems in the early 1980s. SoftCraft, the company that developed Btrieve, was acquired in 1987 by Novell, the vendor of the leading network operating system at the time (NetWare). As a result, Btrieve became a more tightly integrated part of the NetWare OS. Layered capabilities, including NetWare SQL, were developed as layers on top of the Btrieve storage manager.
In 1994, Novell decided to refocus on its core network operating system capabilities, and its database technologies were spun out into a new company, which was renamed Pervasive Software in 1996. Pervasive’s focus is on cost-effective SQL-based databases for use by ISVs and VARs. Packaged software for accounting, inventory control, order processing, and similar functions use it as an underlying bundled database manager. These products are typically sold to small and medium-sized businesses, and to departments of big companies.
Pervasive’s current product, Pervasive SQL, combines their Scalable SQL and Btrieve products. The emphasis is on features important to the small/medium business market. These include low database administration, scalability to support business volumes, a small DBMS footprint, and the capability to handle reasonable data volumes at low cost. Overwhelmingly, Pervasive SQL is used by an ISV or VAR and delivered as a bundled component of their software product, often invisible to the end-user.
18. PointBase (www.pointbase.com)
PointBase is the developer and marketer of the PointBase DBMS, a 100 percent Java SQL-based database. The company was founded in 1998 by Bruce Scott, who had already had a very successful career in the database business. In 1997, Scott was one of the founders of Oracle Corporation, involved in creating the first several major releases of Oracle. In 1984, he cofounded his second successful database company, Gupta Technologies, with its SQLBase product.
The PointBase products are focused on enabling mobile computing, with its special requirements—replication and synchronization of data and providing database capabilities in a very small memory footprint on the client side (e.g., in handheld devices). To serve this market, PointBase comes in three versions. A micro version provides the smallest footprint, and is appropriate for very constrained environments, such as battery-powered handheld devices. An embedded version increases the footprint for systems with a larger memory budget, but where the database is still invisibly embedded in the application. A server version provides the back-end.
19. PostgreSQL (www.postgresql.org)
The Postgres object-relational database traces its roots to the University of California at Berkeley, home of the pioneering Ingres relational database. From the late 1980s to the early 1990s, Professor Michael Stonebreaker and his colleagues worked on extending the relational model to include object-oriented capabilities, resulting in the Postgres prototypes. In the mid-1990s, Stonebreaker used the Postgres foundation as the basis for Illustra, a commercial object-relational product that eventually became the flagship Informix product after Illustra was sold to Informix. In parallel, a group of database experts at Berkeley continued to work on Postgres itself, adding SQL capabilities and distributing it to the research community as PostgreSQL.
With its university roots, Postgres was a natural fit for the open source movement, and began to build its own following as an open source database. PostgreSQL.org is the organization that was eventually formed to organize and coordinate PostgreSQL development. Today, it acts as distributor, support mechanism, and clearinghouse for PostgreSQL distributions, with contributions from a growing user community.
20. Quadbase Systems (www.quadbase.com)
Quadbase-SQL is a SQL-based client/server database system for IBM-compatible PCs. It was originally offered in the early 1990s as a DOS/Windows database with a fileserver architecture. It has since evolved into a client/server database, with support for NetWare, Windows, and Windows NT-based servers. The Quadbase
SQL implementation is ANSI SQL-92 compliant at the Entry level. It provides both Embedded SQL interfaces (for C, C++, and SmallTalk) and an ODBC callable API.
Quadbase supports a number of advanced SQL features including updateable scroll cursors and views. Its multiuser concurrency control offers the flexibility of multiple isolation levels for balancing database integrity requirements with performance concerns. Quadbase also supports read-only schemas that allow it to be used to create and access read-only databases on CD-ROMs. In recent years, the company has placed increased emphasis on its charting and reporting tools, and deemphasized its database management products.
21. Red Brick Systems (See IBM Corporation)
Red Brick (named after the red brick building where the company was founded in Los Gatos, California) was an early pioneer in the data warehousing market. Its founder, Ralph Kimball, remains a recognized expert in data warehousing. The company’s core offering is a SQL-based DBMS that is heavily optimized for data warehousing applications.
Optimizations in the Red Brick system include high-performance data loading, with a parallel loader capability for exploiting SMP systems and high-performance data transformation, cleansing, and integrity checking. The Red Brick software also allows automatic precalculation of aggregate data values (sums, averages, minimum, and maximum values) during the table loading process.
The Red Brick DBMS also focused on a high-performance implementation of the star-schema structure often found in data warehousing applications. Its STARindex technology and associated STARjoin capability implement support for star schemas within the database structure itself. The DBMS also features adaptive bitmap indexing for rapid data selection from very large tables. SQL extensions within the RISQL language handle typical decision support query structures, such as selecting the top three or the 95th percentile of rows based on some numerical measure.
Despite its early lead in the data warehousing market and several early customer successes, Red Brick found its early momentum hard to sustain. Other, much larger database vendors, including Oracle Corporation, Sybase, IBM, and eventually Microsoft, saw data warehousing as a major market opportunity and announced (sometimes with much-delayed shipment) data warehousing capabilities for their product lines. Although its products retained acknowledged technical advantages, Red Brick saw customers decide to wait for their current DBMS vendor. The company was sold to Informix Corporation in 1998, and the Informix database management products were subsequently sold to IBM.
22. Sybase, Inc. (www.sybase.com)
Sybase was a hot mid-1980s DBMS startup company, funded by tens of millions of dollars in venture capital. The company’s founding team and many of its early employees were alumni of other DBMS vendors, and for most of them, Sybase represented the second or third relational DBMS they had built. Sybase quite effectively positioned its product as the relational DBMS for online applications, and stressed the technical and architectural features that distinguished it from contemporary SQL-based DBMS products. These features included the following:
- A client/server architecture, with client software running on Sun and VAX workstations and IBM PCs and the server running on VAX/VMS or Sun systems
- A multithreaded server that handled its own task management and input/output for maximum efficiency
- A programmatic API, instead of the embedded SQL interface used by most other DBMS vendors at the time
- Stored procedures, triggers, and a Transact-SQL dialect that extended SQL into a complete programming language for building substantial parts of an application within the database itself
Aggressive marketing and a first-class roster of venture capital backers gained Sybase the attention of industry analysts, but it was a subsequent OEM deal with Microsoft (the leading PC software vendor) and Ashton-Tate (the leading PC database vendor) that positioned the company as an up-and-coming DBMS vendor. Renamed SQL Server, the Sybase DBMS was ported to OS/2 (at the time, both IBM’s and Microsoft’s strategic future PC operating system) to be marketed to computer systems vendors by Microsoft and through retail computer channels by Ashton-Tate. Sales from the alliance never met early expectations, but it propelled Sybase into the DBMS market as a serious player. Today, SQL Server (several generations later) continues to be Microsoft’s strategic DBMS for Windows NT; Microsoft has split from Sybase, pursuing its own development path. Sybase remains a major DBMS vendor, but the positive impact of its formative alliance with Microsoft has long since passed.
The innovations that made the Sybase product unique in the late 1980s were eventually copied by the other DBMS vendors. Sybase’s early lead cemented its leadership position in market segments that demanded high-performance OLTP, including especially financial services applications. These niches remain Sybase strongholds today. During the 1990s, Sybase expanded its product line to include development tools through a merger with PowerSoft, one of the leading DBMS tools vendors. Other mergers and acquisitions brought consulting services and other data management technologies.
Sybase’s current product line has three distinct database engines, focused on three different segments of the database market:
- Sybase Adaptive Server IQ is focused on data warehousing. It features complex query optimization techniques that are claimed to improve performance by 100 times over conventional RDBMS’s.
- SQL Anywhere is focused on mobile computing. It features a small footprint and integrated support for Java classes and objects as well as Java stored procedures.
- Sybase Adaptive Server Enterprise is the successor to the Sybase SQL Server products, optimized for OLTP workloads. It features flexible locking strategies and query performance improvements.
Together with the Sybase application server, other middleware products, database development tools, messaging products, and consulting services, these product lines make Sybase a multihundred-million-dollar database supplier.
23. TimesTen Performance Software (www.timesten.com)
TimesTen is a venture-backed database company focused on delivering ultra-high- performance in-memory database systems. The company was formed as a spinoff of a main-memory database project at Hewlett-Packard, and its underlying technology has been shipping as an embedded component of HP telecommunications systems since 1996. TimesTen’s version of the technology began shipments in 1998. It features an ODBC and JDBC APIs and industry-standard SQL, and runs on Windows server operating systems, Linux, and UNIX-based servers from HP, Sun Microsystems, and IBM.
The TimesTen in-memory database is targeted at applications with high-performance requirements in telecomm/datacomm systems, high-volume Internet applications, and financial services applications. It has been deployed as a stand-alone data manager within cellular networks and datacomm applications. It has also been used as a high- performance data cache front-ending conventional disk-based RDBMS systems in Internet applications, and as a stand-alone database for stock trading and market data distribution applications
For typical OLTP applications, the TimesTen engine delivers at least ten times (1000 percent) the performance of a fully cached conventional RDBMS. TimesTen 4, the current version, supports 64-bit database addressing, allowing in-memory databases of tens of gigabytes. In addition to its RDBMS features, TimesTen offers N-way data replication capabilities for high-availability and load-sharing configurations and an event-publishing capability. The company’s main-memory database products have been measured at transaction rates exceeding 10 million SQL read operations (read based on primary key) per minute.
24. Versant Corporation (www.versant.com)
Versant was one of the early object database vendors. Its first OODBMS product shipped in September 1990. The current version of its database product offers Java, C++, and SmallTalk interfaces. The object database engine is multisession and multithreaded, and it runs on Windows NT and UNIX platforms. One of its distinguishing characteristics is fault-tolerant capability with automatic failover.
Like all of the pure object database vendors, Versant initially presented itself as a next-generation DBMS system, rejecting the relational vendors and their systems as yesterday’s technology. More recently, the company has opened its OODBMS to the relational world through the Versant SQL suite, providing SQL access and an ODBC API. The SQL facility, and a corresponding Interactive SQL utility, are available for Versant servers on Solaris, AIX, HP-UX, and Windows platforms. A caching version of the Versant product, named Versant enJin, provides an object-oriented data cache for use in conjunction with J2EE-based application servers.
The philosophy of the Versant SQL suite is to automatically present as much of the OODBMS capabilities in a relational model as possible. It automatically maps the Versant database’s object schema to a corresponding SQL schema: for example, it transforms two object classes with many-to-many relationships into two base tables and an intersection table to represent relationships. SQL schema information is available through virtual SYSTABLES, SYSCOLUMNS, and SYSINDEXES catalog views. Embedded pointers within the object schema are exploited transparently to enhance query performance. In addition to the programmatic (ODBC) and interactive SQL interfaces, the SQL suite includes data loading and extraction tools to move information between the Versant OODBMS and conventional RDBMS systems.
Source: Liang Y. Daniel (2013), Introduction to programming with SQL, Pearson; 3rd edition.