Even with many household names in database vanishing, database vendors are not about to give up the fight to host the ever-growing volumes of data being
generated every day.
As soon as the technology was invented to process data faster and more reliably than humans, people needed a way to store their data electronically. Along came the database, and before long the relational database arrived offering faster storage and querying capabilities on large volumes of data than ever before.
As everyone involved in the industry knows, the benefits of relational technology made it the standard database of choice around the world, a leadership position it has held for years even as newer technologies have been developed. Today, relational databases rule the roost, having seen off challenges from other bright ideas, such as object-oriented systems.
Of course the relational systems have changed over the years to incorporate new features and functionality, and to keep ahead of newer technologies. Unfortunately for the relational vendors, these additions have provided competitors with targets. The common argument against relational systems is that the more functionality that is added on, the more cumbersome the technology becomes.
But, says Leon Bouwer, product manager at Bateleur Software: "There is not a chance that we are looking at the end of the relational database. Misperceptions creep in when people fail to realise that different database designs are required for different purposes. For example, nothing beats Software AG's Adabas for online transaction processing (OLTP), just as nothing beats relational structures when it comes to query processing in applications such as business intelligence."
Commenting on whether post-relational, object-oriented, or some other form of database will form the foundation of business data in the future, Bouwer says this will depend on how business needs change and to what degree these database structures end up supporting those business needs. "In my opinion the only problem with relational database is its lack of performance, but it will most probably be some time before a better data structure emerges."
There are however, alternatives to the relational systems on the market today, many of them gaining traction in specific verticals, or offering more attractive partnership options to get developers on their side. Nevertheless, they have scarcely made a dent in the overall database market where the big money is.
Network Times/Computer Business Review Africa asked two competitors in the database industry with different technologies and approaches to the market - InterSystems and Oracle - for their views on the database market and where it is going. Mike Fuller, regional marketing director of InterSystems and Deon Roos, principal sales consultant, database, at Oracle SA offered their input to our database questions.
Mike Fuller, regional marketing director of InterSystems
Deon Roos, principal sales consultant, database, at Oracle SA
The result at the end of the debate is the same as always: relational systems still rule the roost and will for the foreseeable future. Newcomers are delivering the goods with new technologies, but crossing the enormous divide to full acceptance and integration into today's business applications is not something due to happen anytime soon - except in specific verticals where the new technologies can outperform the dominant players.
Q: Are we looking at the end of the relational database?
Mike Fuller: I do not think we are seeing the end of relational databases, but we are seeing the end of their dominance. People are using increasingly varied data types and extended data relationships that often do not efficiently break down into rows and columns. They are seeking alternatives to relational databases not just because their applications are more complex, but because they realise that other database technology options may provide a better fit for what they want to do.
The workarounds required by relational database technology to handle something as prolific as object-based business logic are proving to be less efficient than people are willing to endure. As businesses are consolidating views and processes they question why they need separate often duplicated databases for transactional, data warehouse and analytical operations. Organisations are waking up to the fact that they should expect more from the technology.
Deon Roos: New licence sales for relational database management systems software market grew 10,3% in 2004. Oracle grew by 14,6%, outperforming general RDBMS growth (Gartner May 2005). According to IDC's report, 'Worldwide RDBMS 2004 Vendor Shares: Preliminary results for the top five vendors show a solid boost (March 2005)', Oracle continues to be the overall leader in the worldwide relational and object-relational database management systems software market:
* Oracle Database grew 14,5% year over year, increasing market share lead to 41,3%;
* Oracle outpaced the overall worldwide market for RDBMS which grew by 11,6%.
Those are not the numbers one associates with a dying technology.
Q: Will post-relational, or object-oriented, or even some other form of database form the foundation of business data in place of relational systems?
Mike Fuller: Non-relational databases are out in the market and thrive where they have found niches in which they serve a clear purpose, such as native XML databases. Yet, each of these often suffers from the same problem as the relational databases: they think that one design paradigm can solve many purposes.
The fact is that as database technology evolves to meet new requirements it needs to be reinvented. The main database vendors, Oracle, IBM, and Microsoft, have continued to layer new features on top of their relational technology in stacks that will eventually topple. Oracle state in their own literature that in the conversion of Java object models to their relational persistence there is a 40% design overhead, this is referred to as 'impedance mismatch'. Smaller, innovators have brought to market niche solutions that have had a single-shot use and usually lack interoperability. Neither of these is sufficient for modern computing.
Deon Roos: I would love to predict the future and make a fortune, but at this point in time there is no clear indication that the above will happen - that I am aware of. Unlike non-relational systems, packaged application vendors like SAP, Siebel, PeopleSoft, etc, all rely on Oracle as the underlying trusted database. What this means, is that it would be difficult to add your ERP/CRM application of choice, on to a non-RDBMS. The consequence would be for the customer to purchase non-RDBMS and try and integrate the two databases - an extremely costly exercise.
Q: What are vendors adding into their databases to add value?
Mike Fuller: Based on what customers are asking us, integration is a big issue and an increasing amount of people want harmonious XML and .NET functionality. Perhaps we should not have been surprised, but while .NET got off to a slow start, it is now vying with Java in terms of the number of queries that we receive from our customers and prospects.
Too many people are fooled into thinking that the database war is over, that database innovation has stalled - they are wrong. New compression and synchronisation techniques are vital to the multichannel, low-bandwidth non-determinant distributed computing models required for low-cost pervasive computing. In short, a user wants to be able to reliably perform complex business transactions using common mobile phones with complete wireless mobility and not care how the data got to them nor how it is stored.
Previously, specialist features such as data mining and spatial extensions are now available in the operational data store, so delivering intelligence and visualisation facilities that are driving the value-chain from data, to information, to wisdom for everyone, is available direct to the common desktop.
Perhaps the biggest driver for change in database technology is to manage unstructured data. The disintermediation provided by the Web has both raised users' expectations and reduced their patience for lengthy procedures for data capture and output that afforded us with the opportunity to validate and transform data. They expect traditional text and numbers, handwriting, sound and video to be seamlessly integrated. To do this properly there is a lot of work to be done at the database level. InterSystems has invented its own 'transactional bit map indexes' that boost operational queries by 300% yet present a minimal operational overhead, ideal for boosting multiformat searches.
Deon Roos: All the 'ilities' remain key:
However, if Oracle did not have the vision and added business value to the current and new customer base, we would not have remained the #1 database for so long and also not achieved the market share as above.
Apart from the unsurpassed features above, Oracle introduced Oracle 10g Grid Computing and Automatic Storage Management (ASM) and Flashback features in 9i.
The goal of Oracle's Enterprise Grid is to take new and existing applications, deploy these on low cost commoditised hardware, increasing the quality of service (through virtualisation and provisioning) and managing the enterprise cost effectively. The key for Oracle is 'Do more, know more, spend less'.
Organisations require databases to be more secure - driven by corporate governance with the ability to grow horizontally while lowering management cost. Databases store a company's most precious asset - its data. This data is required to provide an ongoing service, which implies high availability. Businesses also need to store XML data. In addition, Oracle also provides an integrated data warehouse and business intelligence solution.
Microsoft aims to smash database glass ceiling
By Matthew Aslett, ComputerWire
Microsoft is aiming to break IBM and Oracle's grip on the database management systems market by converting unit shipments into revenue with its forthcoming release of the delayed SQL Server 2005.
"We have sort of had this glass ceiling, and bang, the glass ceiling goes," said Andy Lees, Microsoft's corporate vice president of server and tools marketing and solutions, of the capabilities of SQL Server 2005, which is due for release during the week of 7 November.
"In certain types of solutions where you want very high capacity or scalability, there were certain technologies Oracle had that we did not have," he said, citing advanced clustering and log shipping functionality. "Those capabilities are now built into the box."
Microsoft trails behind Oracle and IBM in terms of relational database market share by revenue, with research firm Gartner giving IBM 34,1% share in 2004, Oracle 33,7%, and Microsoft 20,0%.
However, Lees said that in terms of unit shipments, Microsoft sells more than both IBM and Oracle combined, and with SQL Server 2005, the company is looking to convert its volume dominance into increased revenue market share.
Codenamed Yukon, SQL Server 2005 will deliver new scalability, performance (partitioning), security, availability (mirroring and back-up), and integration capabilities, as well as a new price tag, which should also help the company increase database revenue share.
The product is expected to cost just under $25 000 per processor, while its predecessor, SQL Server 2000, cost at least $5000 less. Nevertheless, SQL Server 2005 will still cost less than its nearest competitors.
Microsoft was the fastest growing relational database management vendor in 2004, with revenue up 18%, while Oracle grew 15,6%, and IBM 5,8%, despite the tempting offer of a major release on the horizon, giving Lees confidence that it can grow even faster when SQL Server 2005 hits the market.
"The product is heading towards being five years old. Oracle has had two new products in that time but we are still growing faster than they are," he said, adding that the new release is the result of four years of engineering, development, and testing.