Ready for prime time? – Microsoft Windows NT operating system

Ready for prime time? – Microsoft Windows NT operating system – Product Development

George Stiglich

Are Microsoft Windows NT server-based computers ready for prime-time deployment in intelligent network systems? Many service providers and their suppliers are starting to think so, as both technology developments and simple economics say yes.

Double-digit growth of wireless communications services worldwide is occurring with intense internal pressure for more cost-effective ways to support revenue-generating services. So it’s no wonder that service providers find appealing off-switch service control points (SCPs) that use standard computer hardware.

Today, service providers deploy SCPs on fault-tolerant, highly reliable, high-end computer systems and Unix-based servers. Computer vendors are considering offering service nodes and SCPs based on the Windows NT server operating system with standard Intel hardware as a low-cost alternative for deploying services in emerging markets.

Among the promised payoffs are new revenue opportunities-such as personalized wireless and wireline services-that are not economically feasible with today’s intelligent network platforms.

But before this happens, the platforms must be made “telecommunications-ready” to meet telcos’ reliability and fault recovery requirements.

Network economics Some equipment suppliers predict that widespread global deployment of the next generation intelligent network platforms will begin next year, with overall savings ranging from 20% to 60%. The savings will be derived primarily from lower equipment and software prices associated with volume sales.

Installation is another factor. Compare the network engineering and installation expenses of deploying a large and complex server in a network infrastructure with the Windows NT server plug-and-play installation approach.

With next generation platforms, difficult system setup configurations are done using wizards, and software updates can be provided on a single CD-ROM or even accessed over the Internet. Additional budget benefits come from lower training expenses because next generation platforms use many common PC components familiar to installation and maintenance workers.

Yet cost savings are just part of the profit equation. Less complex equipment designs will allow new services to be developed, deployed and tested in a fraction of the time it takes existing intelligent network platforms. This lets a service provider move from being a “fast follower” to a “selective innovator” of revenue opportunities.

Next generation platforms can extend intelligent networking service coverage to price-sensitive applications in smaller service provider networks (Figure 1). For example, providers can open new markets by allocating personal space on a server for subscribers, allowing them intelligent call control, voice/data message management, location-based services and Internet messaging/telco integration.

The platforms’ deployment in local exchange carrier, competitive access provider and even corporate networks can become an economically viable option. These systems will also provide common platforms for low-end SCP, Internet protocol and computer telephony integration applications. And they can make an attractive intelligent network base for wireless and wireline services, including cellular authentication centers, equipment identity registers, prepaid calling card services and location tracking such as emergency 911.

Providers will be looking for ways to offload processing tasks from expensive switches and fault-tolerant intelligent networking systems to the Windows NT server-based machines, and to distribute applications to extend services closer to subscribers.

The distributed model makes accessing value-added service more affordable. For example, a country such as Chile that has a central distribution point for wireless service can distribute those services to the sparsely populated north and south with a next generation platform solution. Similarly, a service such as prepaid call validation can be distributed and performed locally, instead of tying up a multimillion-dollar switch.

Reliability challenges In the past, mass-marketed, standard computer technology did not have the reliability and performance required to be deployed in public networks. Now, because of the oncoming wave of powerful Intel central processing units, falling memory and storage prices and ultra-high-speed interconnect technologies such as Tandem ServerNet architecture, Windows NT server platforms carry the raw horsepower to support more than 100,000 telco subscribers on a single system. On their own, however, these systems do not meet the reliability required for telecom applications.

Computing platforms for intelligent networking applications must meet stringent availability and fault recovery requirements. When faced with either hardware or software failures, the next generation platform products used in SS7 networks must automatically recover in less than 500 milliseconds without service loss or data corruption. A single equipment outage can cost perhaps millions of dollars in lost revenue.

Today’s intelligent networks use highly fault-tolerant, or high availability, hardware and software technologies that require as little as five minutes of downtime a year. New software upgrades, configuration changes and hardware replacements can be accomplished without taking the network service down.

Without system modifications, standard servers are not suitable for intelligent networking because they can be down as much as 20 hours a year and must be taken off-line for configuration changes and software and hardware upgrades.

With overall system reliability and fault recovery viewed as two of the telecom industry’s most critical needs, providers cannot simply take Windows NT server-based computers off the shelf, load an SS7 stack with applications and put the computers in their networks.

Fortunately, next generation intelligent networking platform technology providers are blazing the trail to develop new techniques to “harden” Windows NT server systems to greatly increase the reliability, scalability and data integrity of these platforms. Advances are being made in processor clustering-especially for SS7 processing-ultra fast node interconnection technologies and the creation of highly reliable hardware.

Building telco-ready products The goal of next generation platform designers is to use emerging software and hardware technology to fill the gap between expensive, but fault-tolerant, high-end intelligent networking platforms, and inexpensive, but not mission-critical, ordinary file servers (Figure 2). To make these systems safe for public networks first requires new engineering designs that leverage the economic strengths of Windows NT server technology and exploit computer system architecture advances found in today’s fault-tolerant network products.

A distributed computing architecture is the key ingredient. From a software perspective, special code-hardening techniques ensure that individual distributed nodes operate together reliably as a single system to support wireless and wireline service applications. This interconnection helps ensure service availability in the event of individual processor failure. Distributed architectures also offer performance scaling, which adds more Windows NT server systems to meet demands for greater computing capacity.

The foundation for building such next generation platforms is computer hardware, including computing processors, data storage and interconnection links. To achieve major cost savings, computing processors used for next generation platforms are normally off-the-shelf servers bought from PC suppliers that mass market the equipment at low-margin prices. To ensure low life-cycle expense, it is best to use a node supplier that offers full after-sale support, as well as a clear upgrade path that takes advantage of ever-accelerating processor and component improvements.

Computing node reliability requires hot-swappable drives that can be replaced without powering down the server, hot-swappable redundant power supplies that can be changed without interrupting service, and redundant network interface cards to ensure reliable communication links. Reliable ECC memory, PCI hot-swappable and redundant cooling fans also contribute to improved reliability.

Data storage used in next generation platforms can be based around the Fibre Channel Arbitrated Loop technology standard to ensure that storage solutions achieve a high level of performance, reliability and scalability. For smaller applications, dual-channel ultrawide SCSI/RAID storage solutions offer practical high-availability alternatives. To achieve high throughput performance and ultra fast fail-over switching times requires a high-speed interconnection capability between individual computing nodes.

Clustering architectures One of the most critical challenges next generation platform designers face is meeting the intelligent networking system requirements for the SS7 protocol. Traditional clustering methods used to enhance reliability of Windows NT server-based computers in enterprise or Internet environments are too slow to meet the stringent 500-millisecond fail-over timing requirement for SS7 or are too expensive.

To meet this requirement and still be cost-effective is crucial to the platform’s success in telecom. One effective approach is to use a special distributed architecture that physically and logically separates network nodes, which are Windows NT servers used for SS7 network connectivity, from service nodes, which are Windows NT server systems used to handle service applications (Figure 3).

Wireless and wireline service applications and their databases reside on the next generation platform service nodes, which link to network nodes via redundant high-speed ServerNet interconnections. One solution is to employ traditional clustering schemes, such as Microsoft’s MSCS, for service nodes to increase the reliability of these service application computing elements.

However, industry research and product development experience show that task management schemes such as a virtual node manager can reside in the network nodes to reassign application-processing tasks if the service nodes fail.

SS7 processing is accomplished in the front-end nodes, along with providing physical interfaces to switches and signal transfer points. In this scenario, a special clustering arrangement that leverages Microsoft’s client/server design is used to distribute SS7 transactional capabilities application port message processing. If any front nodes fail, SS7 tasks are instantly reassigned to other nodes, achieving SS7 availability requirements. The links of a link set may be distributed across a number of front-end nodes that collectively have the same signaling point code.

To enhance overall network administration, a next generation platform manager provides a common command and control function. To further ensure system reliability, multiple copies of the manager software, along with configuration and state information, can reside in several nodes.

Designing, testing and deploying next generation platforms takes significant resources and time, but this effort has been underway in thelabs of several telecom suppliers for about a year and is starting to yield resu lts. The initial intelligent networking platform deployments should be expected by the first half of 1999.

Although it is doubtful that these platforms will quickly replace today’s true fault-tolerant systems, widespread applications requiring low price tags will be made with the platforms around the world during the next decade.

COPYRIGHT 1998 PRIMEDIA Business Magazines & Media Inc. All rights reserved.

COPYRIGHT 2004 Gale Group