To keep the performance of the telephone networks up to par with today's higher performance computers, there has been an effort to replace lower capacity copper lines with higher-speed fiber optic lines. For customers requiring high capacity lines, telephone companies will run fiber to the premises, giving the customer access to the highest transmission speeds available to carriers and networks today. Fiber connections give financial and banking institutions the speed and quality needed for the large quantity of daily transactions, but it is for a high price. Telephone companies bill for usage or capacit, and fiber rental and facilities maintenance are costly. Efficient fiber usage is crucial. That efficiency can be achieved with Dense Wave Division Multiplexing.
DWDM is a tool that network providers currently use to effectively increase network bandwidth and revenues without the additional expense of new fiber facilities See Figure 1. A fundamental driver behind DWDM's mass deployment in long and short haul markets is the exponential increase in internet traffic. The number of internet users is predicted to top 200 million by 1999 while new applications such as high-definition TV, video-on-demand, telemedicine and videoconferencing loom on the horizon. Telcos have turned toward DWDM to keep up with their demand for bandwidth as DWDM promises to increase the power of one single fiber one hundred-fold. It does this by combining multiple 2.5Gbps signals on a single DWDM fiber. As a result of the aggressive introduction of DWDM technology into the market, it is cost effective for private network applications and other corporate networking needs. Sixteen and 32 channel DWDM systems are shipping today with future capacity increases anticipated in the not to distant future, making better disaster recovery today.
At the base of DWDM technology is the simple transmission of light. See figure 2. Imagine a rainbow, each color carrying a message, distinctly separate from the next, traveling through a glass fiber. With DWDM that fiber is capable of transmitting many colors simultaneously. By using different colored lasers, a DWDM can transmit them at once to an output where it is received by a DWDM that filters the channels, distributing the original message. While the light involved is not visible, the principles are the same. For example, through DWDM, 32 optical channels can be used on a single fiber, 16 in each direction. Because of this, instead of one transmit and receive fiber, there are now 16 transmit and receive possibilities. The connectivity increases as a result and the number of rented fibers required goes down by a factor of 32:1. This is a guarantee for saving money considering the fact that fiber rental is an annual recurring cost. Also, each core could previously transmit around 2.5 Gbps of information. By determining that each virtual fiber is capable of transmitting 2.5Gbps, now 80Gbps can be transmitted down the original core.
No information-based institution, including banks, can afford to rely on one single facility for data processing. Any lapse in service could prompt the loss of millions of customers-- and millions of dollars in revenue, making reliability critical for new, high-powered networks. One false move and customers will take their business elsewhere. With such serious consequences, institutions need to have a reliable and speedy backup system.
In order to avoid losses due to disaster, most major banks, financial institutions and corporations have some sort of disaster recovery plan. Such planning has been practiced by financial houses and major corporations for years. Each day, backup tapes from computers are taken to an alternate site for secure storage and are kept in vaults. Some disaster recovery plans include a machine on hot standby at an alternate site that awaits the arrival of tapes in the event of a disaster. With this system, the tapes simply need to be reloaded and communications are re-established to users making things as they were before the disaster occurred.
Backing up data with tapes using a system on hot standby would seem sufficient, but it suffers from considerable drawbacks. The tapes can take several hours to restore and they only contain yesterday's or at best, this morning's data.
Such a time penalty alone to a major clearing bank could lead to losses, as discussed previously, that run into millions of dollars.
The backup system needs to be 'live' hot standby. The machine is literally powered and running and with the flick of a switch can take over from the main machine at a moment's notice. This standby is usually located at a second location to ensure that natural and other disasters do not cause both systems to fail. This is usually implemented by having two identical computer rooms anywhere between 15 and 50Km apart, linked by high speed networking and/or remote peripherals. In the days of slower peripherals, this could have been handled by E1/T1 and E3/T3 TDM Mux equipment.
Because higher speed peripherals are involved and applications are becoming robust enough to support live mirroring on servers, disaster recovery is taking an evolutionary step forward. Backup locations are now looking towards live mirroring of sub-systems ensuring that the data in the remote site is tracking second by second so that data is current and backup systems are already live. Live mirroring involves two hard disks. When a computer's data is written to one hard disk, it is automatically written to a second; if one fails, the other has identical information. This is key for transaction-intensive operations, where it is necessary for a backup system to pickup immediately. The second system backup keeps data accurate to the second, and allows time for technicians to get the original system up and running. The peripheral designers have included high-speed fiber optic interfaces that allow connection of tape drives and hard disks, up to 40Km distant. If one device is required, a single fiber pair will be used, and the system will be complete. As many such devices are connected, multiple fiber pairs are required. As fiber is expensive to install, rent and maintain, efficiency is necessary. For example, Osicom's GigaMux Metro DWDM system allows many optical signals to transverse the same fiber and can transmit as many as 32 duplex pairs, resulting in a line rental saving of 32 to 1.
This example can be seen in figure 3. The two sites, the main and alternate data storage facility, are 10Km apart and consist of four hosts running back to back and Asynchronous Transfer Mode (ATM) switch for backbone traffic.
One particular feature of this implementation is the use of the fault tolerant trunk module, which allows the same signal to be transmitted down two different routes. Fault tolerance makes a system capable of withstanding damage to its infrastructure while still maintaining service. In the event of a trunk failure due to construction work, carelessness or natural disaster, within milliseconds, the receiver will detect the fact that the signal has gone and switch to the backup fiber. This adds a secondary level of protection to the data integrity. Osicom's fault tolerant trunk modules are a pair, one transmitter and one receiver. The transmit splits the signal and sends it down two fiber cables, each of which should be separately routed. Both fibers come together again at the receive Gigamux that monitors both. One is designated as the working or primary fiber and the second is the protection fiber. If something should happen to the working fiber, the protection fiber is ready to step in and take over.
This feature is demonstrated in figure 4. Two computer centers are 17Km apart, and the Gigamux is bused to provide virtual fibers for peripheral interconnect. Initially this is used to ensure second by second tracking of data. It provides security for the institutions involved and a backup if any disaster should occur.
As electronic banking becomes more popular, banks will have an increased amount of data to transmit and store. The loss of this data in the event of a disaster would cause millions of dollars to be lost. Because the consequences are so severe, banks plan for disasters by storing backup information at an alternate site. Because these sites are generally far apart, fiber technology is used to connect the storage facility with the transaction facility. Using DWDM provides significant savings in the cost of fiber rental. As DWDM products create 'virtual fibers,' many sites requiring multiple fiber interconnect are good candidates for its use. DWDM also can provide a successful fault tolerant plan to allow for flawless data communication.
Mark DiMaria has been with Osicom since 1997, where he heads up product marketing and management for the GigaMux Metro DWDM product family and the Network Adapter cards for FDDI and Ethernet. DiMaria's expertise is in Internet backbone technology, DWDM technology and application, and data network design. DiMaria can be reached at mdimaria@osicom.com .
Neil Grayson joined Osicom Technologies in 1996 to head up engineering for the company's European operations. He has over 20 years experience in designing and developing hardware systems, both in Europe and the United States. Grayson can be reached at ngrayson@osicom.com . For company, product or financial information, visit the Osicom website, www.osicom.com .
