Timing is EverythingNetwork performance relies on the network of clocks that provide timing and synchronization. by Austin Lesea The phone rings at 3 a.m.: a customer reports that he is taking excessive bipolar violations, CRC-6 errors and other impairments. The network is down, and your customer is "out of business." As the service provider, how do you find the problem, and how do you go about fixing the problem once you've found it? More important, how do you prevent this type of problem from happening again? Impairments like the previous example have many causes. Increasingly, the problems relate to synchronization quality--or lack of quality--for the various network elements. Synchronous Optical Network (Sonet) transmission and switching systems require proper synchronization. Take a look at a typical central office or timing node: It usually contains a digital access cross-connect switch, part of a Sonet fiber ring, perhaps some channel banks, maybe a Class 5 local switch and, on occasion, a Signaling System 7 signal transfer point (STP) processor for billing. Each device requires a high-quality synchronization signal that can be traced to the Stratum 1 (primary reference) level timing source in order for the equipment to operate properly. (Stratum 1 is a completely autonomous source of timing that has no input for tracking. The usual source of Stratum 1 timing is an atomic standard or reference oscillator. Many of these offices or nodes are collected into a larger network-all interconnected, and many connected to other independent networks as well.) Each connection point become an opportunity for a problem, either causing difficulties in the next node or creating problems even further downstream. BIT BY BITS The building integrated timing system (BITS) concept has been used extensively to synchronize the network. In BITS, one place in the office contains the equipment to supply timing signals to every other element requiring timing. This timing system gets its reference from another location, from Loran C in older systems, or from a global positioning system (GPS)-derived reference for modern systems.
Because every location runs at nearly the same frequency, all locations are syntonous-not synchronous-with each other. Synchronous implies that everything is in phase, which it is not, nor does it have to be. Every independent network has its own primary reference or references, so the system consists of many clocks running at nearly exactly the same frequency. This is known as isochronous. The Stratum 1 American National Standards Institute (ANSI) definition is an absolute accuracy of 1 x l0-11. The complexity of a BITS synchronization network can be overwhelming. Timing information is distributed to offices through a hierarchical series of levels starting with the primary reference source (Stratum 1). The network is designed so that a clock always receives timing from a clock of equal or higher Stratum level. This ensures that if an upstream clock enters into a "hold" mode, the downstream clocks will be able to track it. The BITS clock also may provide primary and secondary signals. However, it is the inherent complexity of BITS that often leads to the infamous 3 a.m. phone call. When something goes wrong with the BITS or in one of the connected elements, the challenge is how to locate and solve the problem. A number of different approaches have been implemented to try and safeguard the network; to date, most of these methods have proven to be costly or marginally effective, or both. Dedicated Tl backhaul. For example, one network provider may decide that, to ensure a high level of service quality and availability, it will route back to a central location a Tl circuit from each of its offices. All of the Tl's then would be analyzed at that central location for any timing impairments. By building this system, technicians could identify which office was not in sync.
This system works somewhat, but it may not be viable from an economic and performance viewpoint. With today's trend to build many more satellite offices and distributed networks, the cost of Tl's becomes prohibitive (in direct costs and cost of lost business), for these Tl's could be producing income otherwise. An additional challenge is how to determine if the problem is an office one or a transmission one. Inquiry system. Another network provider has the ability to call up central locations responsible for the primary reference. By "talking" to these locations, then "talking" to the network elements that have information interfaces (such as a DACS), the network provider can identify problems down to the office level. The drawbacks are that not every node has a DACS, and there are many lower-level BITS that are expected to perform their jobs without failure. In other words, there are holes in how the network is managed. Head in the sand. A more common solution is the head-in-the-sand approach. Here, when a problem arises, the solution involves gathering various test sets, a cesium Stratum 1 standard, and test technicians and engineers, and sending everything into the field. But where are they going? At this point, it’s not certain where the failure has occurred. The head-in-the-sand approach often equates to long system outages, because of the time involved in locating the problem site and identifying the faulty equipment. It is unusual for the technical team to travel to City A, only to find out the problem is not at City A and requires additional travel to City B. Often, the team, upon identifying the exact problem, discovers that the situation could have been resolved by local craft technicians in a manner of minutes, had a system capable of fault isolation or identification been in place. INTEGRATED MANAGEMENT A recently introduced integrated network management system holds the promise of providing a cost-effective solution for pinpointing the location of a network timing problem and identifying the source of the trouble. The capability to communicate with each location and identify the conditions and the status of the alarm permits technicians to isolate and quickly identify the exact nature of the problem. In most instances, the problem can be resolved by onsite technicians, usually without any system downtime. The integrated management system gives each element in the network the ability to report its status to a centralized network the ability to report its status to a centralized network management location. Now the question becomes, What do you need to know? To answer that, take a step back and look at what you would like to have known when you received that call at 3 a.m. Each T1 circuit has various performance parameters that are useful:
In a perfect world, there would be an agreed-upon set of service thresholds. If any of the above parameters exceeded those preset thresholds, an alarm would be generated and set to a network management center, or logged locally, or both. This is the information model for a system. In this case, it is the information model for the timing system. The International Telecommunication Union is considering what a timing system information model should look like. The submission covers all of the above items, as well as internal parameters of the timing system itself. By combining these information models for the various network elements, and having a common interface language, a network management center (or a technician using a laptop computer from his home) can interrogate the network and determine the cause of the problem. By establishing a performance model for the network and using the integrated network management system, the service provider can detect a problem when a network crosses its threshold. This allows the provider to pinpoint the location and source of the problem immediately. Moreover, it permits the problem to be solved before that customer goes "out of business." The digital switched network has become the lifeblood of telecommunications. The failure of any element of timing within a given network can lead to service failure which, in turn, is costly to the consumer and to the provider. Previous attempts to establish precise and instant timing problem-solving throughout the network have been costly or ineffective, or both. The advent of the ITU-T common language network performance models, coupled with an integrated test and measurement system, show great promise in bringing together the elements required to provide a much-needed centralized monitor and alarm capability to the telecommunications industry. Austin Lesea is Vice President of Advanced Product Development and a Director of Larus Corp., San Jose, California ©Reprinted with permission from AMERICA'S NETWORK, September 1, 1996 AN ADVANSTAR PUBLICATION Printed in the U.S.A.
Home | Timing Tutorial | T1 Tutorial | Larus Articles | Client Data Access
|
 |
 |
 |
||||
| ||||||
|
