JDN et all, good article as to why RDBMS are dead,and OBDMS are gonna rock,scalability.
Distributed object-oriented databases
Mapping to the converging networks services
MIKE GURLEY
Value-added services represent an exceptionally attractive source of additional high-margin revenue to network-based service providers. Today's network service providers must deal with the issue of delivering new value-added services from a converging network architecture composed of wireline, wireless, Internet and cable networks.
The latent demand for these services is extraordinarily high, particularly considering how few have actually been deployed. As each new capability hits the market, be it voice mail, unified messaging, digital wireless telephony, Internet telephony, electronic commerce or the like, it is immediately pounced upon by customers, resulting in a dramatic infusion of new money into the marketplace. That money is going to the few companies that have been able to deploy systems and networks rapidly and effectively.
The top-level conclusions here are straightforward:
People love to communicate and are ready to invest significant additional funds to extend their range of communication, and
The only significant barrier to massive market growth is the development and deployment capabilities of the network-based service providers.
A less obvious conclusion is the transfer of power happening in the telecommunications industry from established providers to those technologically nimble enough to claim the new prizes with all of the forces driving the network landscape.
We have seen this in the way Metropolitan Fiber has successfully scooped out the prime local loop business in larger cities. We also have seen it in the way MCI has created havoc in AT&T's long-distance market with its "Friends and Family" promotions. And we are going to see the same trend in value-added services.
The question is: Where will your company be when the dust settles?
The Technology Issue
The technology at stake here falls within an increasingly well-established network architecture model called the Intelligent Network. The IN was designed to enable value-added services to be developed and deployed outboard of the main switch.
Leveraging commercial UNIX-based computers from companies like HP, Sun and others, this model provides service control points (SCPs) and analogous platforms (adjunct processors and intelligent peripherals) as shared or dedicated devices for deploying value-added services. The operational supports systems (OSSs) must also be in sync with the availability of any new service.
These new services take traditional telecommunications capabilities and data sources and combine them with the capability of the Internet in innovative ways to create new value for customers. Personal phone number, voice mail, voice over IP, virtual private networks (Internet based), call forwarding and call waiting are just a few examples, behind which lurks a virtually infinite, untapped and lucrative field of opportunity (Figure 1).
To implement these innovations, software providers have turned almost universally to object-oriented software development techniques using languages like JAVA and UNIX C++. Compared to traditional software development, these techniques have numerous advantages, two of which are critical to the value-added services domain:
They have an event-response bias that is inherently suited to handling dynamic processes like phone calls, faxes, etc. (where as traditional techniques have a command-executive bias that is suited to the automation of prescheduled processes); and,
They have a peer-to-peer bias that is best suited to handling the many-to-many relationship interdependencies involved in all these services (whereas traditional techniques have a hierarchical bias better suited to handling many-to-one relationships).
For these reasons and others object-oriented software is considered the norm for this industry. There is, however, one issue that is still under debate: What database will be used to support these object-oriented applications?
Relational vs. Object-oriented
The mainstream category of database for worldwide commerce today is the relational database (RDBMS). During the 1980s RDBMSs assumed the leadership mantle from the first generation of hierarchical databases.
The main advantage they brought to the marketplace was much easier access to information in a format that managers understood-- reports, tables and graphs. Most recently, RDBMSs--hooked to personal computers acting as clients--have become the backbone of the current explosion in client-server computing.
For all their strengths, however, relational databases are not well suited to supporting object-oriented applications. Nonetheless, they can be made to fit, and because of developer familiarity with the technology and the market clout of the top companies in this industry--notably Oracle, Sybase, and Informix-- they are still often used. This usage, however, is at the heart of the telecommunications industry's frustration with slow deployment of value-added services.
The problem with relational databases in this context is one of performance and scalability. Because the architectural fit is forced, it is impossible to tune these systems to achieve the processing speeds necessary to support cost-effective commercial deployment.
Java Solutions
in Telco Networks
There is simply too much overhead in maintaining the relational paradigm in an application environment where the paradigm works against the application. Moreover, as the number of customers increases, one cannot scale up these systems to meet the increases in demand, because the problems only get worse as the size of the databases get larger. In short, one can pilot new service delivery systems quickly with relational databases, but one can never deploy them broadly.
By contrast, object-oriented databases (ODBMSs) are specifically designed to support object-oriented applications and simply do not have these problems. They outperform relational databases dramatically--all of their architecture works for the applications, not against it--and those that have a distributed architecture can readily scale up to meet the commercial deployment demand. Why then are they not more in use?
Scouts or Pioneers?
ODBMSs are still relatively new. The technology has been in play for more than a decade, but to date there are limited number of players in the business, the largest of which is only about $50 million in revenue.
Moreover, the products themselves lack some of the features that IT professionals expect and that value-added service applications need. To take on a new database at the same time one is taking on all the other challenges of new application development is a lot to ask.
Finally, none of these database companies focuses only on the telecommunications industry. All of them support a diverse base of customers from manufacturing to finance to defense. As a result, the vendor may be a more fragmented resource than it first appears.
Despite these issues, however, we must return to one fundamental point: ODBMSs represent the only reasonable path forward for new network-based services application deployment, both the service itself and its associated OSSs.
A Game Plan
The total ODBMS market revenue this year is estimated to be $125 million. To date, there has been some noteworthy "early adopter" successes within telecommunications applications.
A major network equipment provider is replacing three mainframes with 5000 workstations worldwide to deploy a configuration management application, all based on an ODBMS from Objectivity, Inc. Another telco recently deployed a network provisioning application to multiple sites, after it replaced the application database with the same ODBMS to achieve the required performance. New code division multiple access (CDMA) and time division multiple access (TDMA) wireless equipment providers are using ODBMSs as the basis for their operational and maintenance and provisioning systems (OAM&P).
Motorola's Iridium System--which began providing global communications on Nov. 1, 1998--also is using an ODBMS as the data repository for system component naming, satellite mission planning data and orbital management data.
Partnerships
Telecommunications vendors and network operators agree on a couple of issues that require significant additional R&D investment:
Very High Availability
Today, fault tolerance is achieved through hardware solutions that put additional demands on new services development teams rather than provide:
Online upgrades to software releases and data;
Replicated data (for supporting remote hot standby sites); and
Safe transaction storage and roll-forward log.
Scalable Performance
A core engine that further reduces transaction overhead and that scales linearly with database size and complexity.
In conclusion, increased revenues go to the companies that are able to deploy systems and networks rapidly and effectively. Object-oriented databases help companies move into new markets with new services--using reliable, high-performance technology.
Mike Gurley is the North American Sales Director for Objectivity, Inc., Mountain View, Calif. His e-mail address is mike.gurley@objectivity.com.
How far along is MAPX in OBDMS,anyone?
Hiram
|
