Optical matrix switches dynamically access network resources

Feb 1st, 1996

Optical matrix switches dynamically access network resources

By linking data and network management platforms to form a mesh architecture, optical matrix switches render full fiber-optic technology access, flexibility, management and control attributes

Carl g. symons and kenneth j. garrett

Astarte fiber networks inc.

Optical matrix switching products offer all the advantages of existing optical switches and provide advanced fiber-optic network management and control features. These products are, in fact, networking systems and should be evaluated when the network architecture is being designed rather than classified as fiber-optic components to be added later.

In operation, they optically crossconnect transmissions to allow multiple wavelength throughput, bidirectional transmission and high-bandwidth transport capabilities. These systems are currently operating in networks carrying digital transmissions from 10 megabits per second to 40 gigabits per second and analog transmissions to 20 gigahertz.

Optical data streams are switched optomechanically using an advanced servo-control alignment process. Within milliseconds, an optical matrix switch can switch a single fiber or dual fibers (transmit and receive pairs) up to 144 fibers. Each matrix switch can be programmed to reconfigure a network hourly, daily or as needed through commercially available simple network management protocol, or SNMP, or Windows network management packages. Optical matrix switches are installed and manage data networks at companies as diverse as Bear, Stearns & Co., the Chicago Mercantile Exchange, Boeing, Comdisco and Bell South Corp.

Switching element

The full benefits of fiber-optic technology emerge when optical switching is used as a network element. When first introduced, optical switches were used to access backup facilities or to test fibers within a cable. They did not, however, directly address networking issues associated with workgroup productivity and were not active elements in network management.

Recovery from fiber-optic network failures, optical testing and reconfiguration are addressed by manual patch panels or by reprogramming the electronic hubs, routers and bridges in data communications networks. This approach works well in small installations, but is not effective for network management in large or complex fiber-optic networks.

The use of fiber-optic technology for data communications networks is being accelerated by decreasing fiber costs, additional users, increased capacity and speed, enterprise-wide applications and bandwidth-intensive video and image transmissions. In turn, these factors are spurring user demands for more dynamic and flexible fiber-optic network architectures.

Consequently, fiber-optic networks must accommodate changes within a company`s business environment and allow the deployment of new network technologies while improving enterprise productivity and profitability. To this end, optical matrix switching provides a means for network users to realize the full benefits of fiber optics technology.

Optical fibers have been traditionally viewed as nonintelligent conduits capable of carrying unlimited amounts of information from point to point. In this view, information routing is accomplished through complicated multiplexing and electronic signal conversions. Shared network architectures such as fiber distributed data interface, or FDDI, and synchronous optical network, or Sonet, expand the utility of fiber optics by giving multiple users the ability to access multiple information streams carried over a single fiber.

These architectures result in the improved overall processing capability of computers, workstations and other elements linked to high-speed fiber-optic networks. Although these architectures can increase network productivity, some applications involving multiple users, multiple networks or high-bandwidth requirements can fall short of fully meeting users` needs employing conventional fiber-optic network approaches.

Fiber-optic routing and addressing schemes depend on the capabilities of electronic protocols and switching mechanisms to distribute information to appropriate destinations. Recent advances in network protocols such as asynchronous transfer mode, or ATM, and Fibre Channel can improve transmission throughput, but these technologies do not take full advantage of the capabilities of fiber-optic networks because they are generally limited by the attached routing devices.

Network reconfiguration

In data communications applications, fiber-optic systems are growing larger and more complex as users demand more bandwidth. As a result, companies are changing the way they use and manage optical data networks. Many companies are using fiber to implement enterprise systems connection, FDDI and ATM networks, as well as low-speed networking schemes such as Ethernet, within the same campus environment. Multiple networking protocols and their dedicated routing systems create network management problems when networks fail or when users need connections to other networks or resources outside their current domain.

For example, an engineering organization working on a large, complicated design project is using optical matrix switches to increase network productivity and decrease downtime. The project calls for separate fiber-optic networks and multiple levels of users, such as systems programmers and hardware design engineers, who require access to a variety of remote databases, simulation programs and field-test operations. Each user is on either an Ethernet or FDDI network carried on fiber. Workgroups are defined based upon individual project resource requirements and specific areas of expertise. Each workgroup ranges from 10 to 30 users and must access as many as 12 different backbone networks.

Several alternatives were evaluated in developing a network topology that would optimize the productivity of the entire workgroup. These choices included

Using one large network to connect all potential users and resources

Installing new fiber-optic cable and manually patching each user to a specific resource

Developing a dynamically switched mesh network architecture to give each user access to networks and resources as necessary.

The first alternative was dismissed because of insufficient bandwidth, file transfer delays and protocol compatibility issues. The second alternative proved costly in terms of equipment, cable management and productivity. The optically switched mesh network approach was chosen because it provided the capabilities to reconfigure network resources on demand, to use existing network equipment and transmission protocols, and to manage the project from a centralized network management location.

Existing network configurations continue to operate as usual, serving the needs of their primary users. When a new user, workgroup or network must be accessed, optical matrix switching is used to reconfigure the network and establish the required point-to-point connections. Because fiber is the common transmission medium for all users in the project, network control is managed at the physical layer and does not interfere with the complex networking algorithms associated with electronic routing.

Performance management

Optical matrix switching can also play an important role in network management. When combined with simple network management protocol- or Windows-based management platforms, optical switching systems provide effective means to test and monitor network availability and performance. Optical matrix switching can work with detailed network management software programs such as Hewlett-Packard`s Openview and IBM`s Netview, while providing an end-to-end optical network architecture.

An important area of productivity improvement and cost reduction using optical matrix switching systems involves the reduction of network downtime. Without matrix switching, network managers generally depend upon a single recovery path. An optical matrix switch offers the ability to route transmissions to multiple preprogrammed recovery paths based upon available network resources. This approach allows effective traffic management within the network and eliminates concern about the readiness of dedicated backup systems. Once a recovery route has been established, diagnosis and repair become simplified through the system`s ability to directly access problem fibers, perform optical loopbacks for fault isolation and remotely insert test and monitoring systems.

Two types of network testing procedures are commonly performed in data communications architectures. The first test involves checking the integrity of the physical cable through the use of an optical time-domain reflectometer, or OTDR. These instruments can provide a network manager with information on the performance of the optical path, but they do not provide information on specific data transmission characteristics. The second test involves the use of a local area network or ATM analyzer, which supplies data stream performance characteristics.

In most data management centers, networks are tested manually. An OTDR or local area network/ATM analyzer is plugged into the network at a router or patch panel to perform testing. Large, complex fiber-optic networks or campus environments containing numerous router and hub locations, therefore, require multiple test instruments or the relocation of test equipment as needed. Both test setups are costly.

To improve productivity and reduce network management costs, data center managers are consolidating their control systems and implementing remote monitoring capabilities. Optical matrix switching allows network managers to remotely access any fiber or group of fibers without the help of technicians or mobile test equipment. In addition, remote access through optical matrix switching eliminates the need for multiple test systems, which saves costs.

For example, an ATM analyzer equipped for fiber-optic transmissions can cost as much as $500,000. A major brokerage company in New York City is currently upgrading its network to ATM and recognized the need to effectively manage and test its network. By installing optical matrix switches, the company can use one analyzer to monitor and check several locations and effectively manage its network.

Physical network management

Cable management and recabling are expensive items in a data networking budget. The more dynamic and volatile a company`s environment, the higher the cost to provide network resources to meet operational requirements. To meet the anticipated needs of users, network planners are prewiring new buildings and campuses with fiber-optic cables. In this way, they save costs by completely cabling a building or campus during construction rather than rerouting or installing fiber-optic cables later. The management of these fiber facilities, however, calls for an optical matrix system that can efficiently link users, networks and resources.

Optical networks must dynamically respond to changes within a company`s infrastructure. Entire departments, workgroups or individual users are continually moving within a building or to other buildings. Optical matrix switches combined with network management software platforms allow a fiber assignment procedure similar to that used to assign telephone numbers within a company`s telephone system. In this approach, network managers can track fiber assignments to a specific device or user level. When a user or device is moved to a new location or reassigned to another network, the new topology can be quickly and easily configured without impacting other users or networks. The optical matrix switch allows the network control center to keep pace with personnel and workgroup assignment changes. u

Carl G. Symons is president and chief executive and Kenneth J. Garrett is vice president for sales and marketing at Astarte Fiber Networks Inc. in Boulder, CO.

How Optical Matrix Switches Work

An optical matrix switch contains a group of input fibers and a group of output fibers. The optical signal from each input fiber is focused onto a collimated light beam passing through a lens. The beam is then electronically directed to the selected output fiber, where a receiving lens focuses the light signal into the receiving output fiber core. Signals are switched within 50 milliseconds. The number of light beams at any time equals the number of active communication paths. Although light beams from different input fibers can interact, no crosstalk is incurred among the intersecting beams.

Servo-control mechanisms are used to automatically control and monitor all aspects of the switching process. The servo-control system eliminates the need for manual system calibration and optical alignment procedures. It continually checks all the fiber connections to detect and eradicate outside influences from vibration and temperature fluctuations. In addition, both sides of the matrix switch are operated by the servo-control system, which in turn, provides switch alignment information to the main processors. Switching commands can be entered by a line terminal, a DOS- or Windows-based personal computer software program or a network management program such as simple network management protocol or Hewlett-Packard`s Openview.

More in Design & Manufacturing