Managing dwdm systems in the metropolitan market
Network-management systems for metropolitan applications will need to meet the information needs of both carriers and customers while providing a migration path to more powerful strategies.
Ron Mackey Osicom Technologies Inc.
The number of Internet users is predicted to top 200 million by 1999, while new applications such as high-definition television, video-on-demand, telemedicine, and videoconferencing loom on the horizon. Even though such demands on the network are growing exponentially, the consumer will continue to demand a high level of service. As the surge of traffic swamps Synchronous Optical Network (sonet) technology, telephone companies must turn to new technology to keep up with this demand for more bandwidth and high service quality.
Metropolitan carriers are turning to dense wavelength-division multiplexing (dwdm) to meet these needs because it promises to increase the power of a single fiber 100-fold (see Fig. 1). Given the projected universal deployment of dwdm, one of the largest issues users and providers face is how dwdm networks will be managed and monitored.
The melting pot
Typically, metropolitan centers are characterized by a concentration of business, governmental, and educational organizations, each representing a user type that responds differently to new applications and service features. This scenario contrasts with the typical long-haul applications in which dwdm previously has been applied. In today`s metropolitan network, if a new application is competitively priced, it can trigger an instant and unpredictable demand for additional bandwidth because a particular demographic takes an interest. Because this shift can happen so quickly, reliability is a critical factor in new high-powered networks. Industry deregulation also has fueled the pressure to maintain a highly reliable network by creating an array of carrier choices for the end user.
Not only has deregulation stepped up industry competition, it also has presented great challenges for maintaining network integrity because of increased network diversity. Under the old Bell system, networks existed under a single carrier, making them homogeneous. The current offerings from equipment suppliers found in different networks use an array of protocols, presenting a challenge for any new network-management technology. Generally, many local/Internet users remain focused on network-management systems based on Bellcore or International Telecommunications Union standards. In the corporate world, as well as in intranet/Internet services, snmp (simple network management protocol) is clearly the dominant choice. Carriers have opted predominantly for TL-1 (transaction language 1) and may choose the upcoming Q3 (cmip/cmis--common management information protocol/common management information services) or in the distant future corba (Common Object Request Broker Architecture) protocols. This lack of a single standard makes shared management difficult to achieve.
The financial success of long-haul dwdm vendors testifies to the technology`s maturity and effectiveness in multiplying existing fiber activity while keeping infrastructure costs low. Just as the bandwidth squeeze prompted long-distance carriers to turn to dwdm, it is also a viable solution to alleviate network pressure in the metropolitan arena. However, if dwdm is to be successful in the metropolitan market, it`s vital that it function with a multiple-access-management scheme that everyone can use--users, craftspeople, and operators alike.
Channel transparency for success
In the past, network operators provided links to customers over several independent fibers. dwdm will mix several customer links onto a common fiber (see Fig. 2). This creates a challenge for a metropolitan network-management scheme, because it must accommodate all those with a need to know management information--capacity providers, wholesalers, and large end users. The mix of broadband applications and services--data, voice, or video over Digital Subscriber Loop; 100-Mbit/sec Fast Ethernet; Gigabit Ethernet; Asynchronous Transfer Mode; frame relay; or sonet--dictates that dwdm systems be channel-transparent to any data format or bit rate and adds to the complexity of network management.
Having many links on a common fiber creates an additional concern: security. Management systems must maintain the same security and isolation between different customers that is provided by transport over different fibers.
Overall, the chosen metropolitan system must be totally channel-transparent and provide standardized network-management systems, as well as a migration path to next-generation management architectures (see Fig. 3). Because the dwdm system will lie at the heart of the network, metropolitan network-management systems also must be interoperable and have the ability to ultimately manage the network end-to-end.
Migration to tmn carrier management
Large operators worldwide have embraced Telecommunications Management Network (tmn), a standard that defines new protocols and a hierarchy of intercommunicating management systems. While TL-1 remains the current protocol of choice in North America at the network element layer of this hierarchy, tmn is emerging as a new solution (see Lightwave, March 1998, page 44). Moving to tmn is not unlike moving from a basic subcompact car to a loaded sport utility vehicle. It has the capabilities to integrate all of the equipment and systems of networks from the past, present, and as far as we know, the future.
The secret to creating a useful network is to have it function without a single point of failure in the management systems. In a tmn scheme, the end devices and network elements in many cases will be managed by an element management system (see Fig. 4). This system communicates upward, or "northbound," in the hierarchy to a network-level network-management station. Here, the information is collected and reduced to depict the status of the entire network and its variety of equipment components. It reports upward again to two more tiers of management systems that function as service- and business-related systems to control important functions such as billing and intercommunications with other carriers.
For the purposes of device management, tmn favors a new protocol between the network elements and the first-tier network-management station. While it doesn`t preclude using snmp between the devices and their manager, tmn clearly recommends Q3, an Open System Interconnection-based protocol. This protocol uses a management information base written in asn.1 (abstract syntax notification) format, which is similar to snmp but much more powerful in the number and type of commands that can be used. For complex products such as large switches and master sonet multiplexers functioning as gateways to their slave systems on a network ring, Q3 is ideal because of its ability to transfer large blocks of information more efficiently and with better assurance of delivery. This ability can be important, as tmn`s extensive list of features can prove cumbersome if not handled efficiently.
The need to know
Given the high cost of high-bandwidth transmissions, end users have a vested interest in knowing what is happening on their links. In general, they need to know how many bits are being transmitted, how many are being received, and how long it takes to get from one end of the network to another, prioritizing traffic according to quality-of-service issues. Specifically, the end user is most concerned with what is being transmitted over the network.
In contrast, service providers are more concerned with how the service is performing. For instance, they must control the physical links that compose the backbone of the network and interconnect the various users. It`s essential for service providers to know the status of their network to ensure optimum performance, facilitate maintenance procedures, and allocate costs. They also need to know the conditions of any channels leased from other carriers.
The challenge of a workable management system is to grant the information required while securing the equipment from unauthorized access.
Since both service providers and end users have embraced tcp/ip (transmission control protocol/Internet protocol) for Internet-related applications, the development of a maintenance management network based on tcp/ip for shared use would allow common access to "need-to-know" information. A cost-effective means of implementing this approach is to install a small 56K integral router and a digital data service copper line from each end site to the provider`s head site (see Fig. 5). This scenario is achievable because IP hubs and routers, as well as digital service units, are now low-cost commodities offering user familiarity with device setup and testing. This simple approach would do much to strengthen relationships between end users and providers. For service providers using dwdm, implementation would proceed as follows.
Customer premise equipment management. At the customer premises, create an isolated Ethernet local area network (lan) segment with provider-supplied IP addresses for each dwdm channel that has been assigned to the customer. The end user can then attach an existing or separate snmp workstation to the lan segment, achieving local management connectivity with the maintenance network. The network provider would supply unique snmp read and write passwords only for the channels that terminate to the customer.
The single rule to enforce is that the network-management system attached to this lan must get its address from the provider and be isolated from the customer`s network, either by separate lan card or physical PC. Other equipment at the customer site should have addresses and snmp community strings assigned for security. To prevent one user from inadvertently disrupting service on a common line, shared items such as optical amplifiers may be monitored by issuing end users a "read-only" community password, while reserving the write password for the provider.
Carrier end-site management. The customer premises equipment can be connected to a carrier-operated lan via the dedicated 56K router links. The benefit of using low-cost combination router/channel service unit links is their reliability. They function as an independent route to provide two-way maintenance data links that both the end users and carriers can employ regardless of the operational status of the fiber. Both IP subnetting at the remote-management lan and unnumbered link routing will help to conserve addresses. Routing Internet protocol v2 is sufficient for the routing protocol at these locations.
Customers can access and manage any dwdm channels in the end-user zone for which they possess the appropriate addresses and snmp community strings. Alternatively, the carrier can use snmp to manage everything in this zone as an additional service. In any case, carriers want a high-level view of the network, including all equipment and facilities, and need to see the status data on their existing network-management systems. In contrast, end users are interested in a view of their circuits that will allow them to be monitored for performance and availability.
Zone-to-carrier network management. A single or dual PC arrangement can provide message conversion for a large number of devices into the carrier`s native management protocol. Which protocol to use remains an issue, since either TL-1 or Q3, depending upon whom you talk to, is the desired choice. The PC offers a good migration strategy for carriers who plan to move to tmn Q3, because the difference between the protocols is a matter of software.
To fully convert all parameters from snmp to other protocols, an active database of configuration and status would exist in the PC. Changes in this adapter software would be synchronized with the network provider`s main management systems migration. The benefit is that the actual equipment distributed around the city would not have to change at all, and the end users could continue using snmp regardless of the carrier`s choice of management protocols. Additionally, the PC could function as a Web server to provide end users with additional information as a supplemental revenue service. Secure Web-server features also could allow individual customers to receive customized reports. Remote-control software would allow authorized users to access information from the PC.
Zone-to-zone access. Within the provider`s network, routers can tie the zones together using isolated addresses. Typically, DS-1 (1.544-Mbit/sec) links would be sufficient between zones. End users could access their remote-end equipment via snmp using these routes. The only requirements are private addresses, since no traffic traverses the Internet. These networked zones also can be used by the carriers to concentrate more devices into fewer network element managers. The physical isolation of this maintenance network from the Internet simplifies many addressing and security issues.
This zone-to-zone management scheme satisfies the critical information status requirements of service provider and end user alike. It also supports and allows the successful adoption of dwdm in the short-haul metropolitan market.
dwdm`s success in the metropolitan area is dependent upon the service providers` ability to identify vendors who can offer dwdm technology and who are skilled at building and deploying management networks. Only in this way can the needs of both end users and network providers be met.
dwdm to the shorter-range metropolitan market promises new high-bandwidth solutions and cost savings for many companies. The burden is on dwdm vendors to offer systems that address the unique demands of metropolitan network operators--systems that are scalable, transparent, flexible, have open architectures, and are equipped to support a flexible management scheme to support the myriad of new customer demands that will surface as consumers take advantage of the emerging technologies. u
Ron Mackey is executive vice president of technology at Osicom Technologies Inc. (Santa Monica, CA). He can be contacted via email at: [email protected]