FrameRelay technology.

FrameRelay technology.

FrameRelay technology

Background of development In the late 80s, one of the few data transmission methods that fully complied with the seven-level BOS system was the X.25 protocol. This technology was created to ensure good performance of user applications during data transmission, regardless of the type of system or equipment manufacturer.

It should be noted that the standard provided for data transmission over channels with a high error rate (BER (Bit Error Rate) = 10-5). For reliable and trustworthy transmission of information, error detection and correction mechanisms, as well as data flow control procedures, were developed and standardized. Such an approach determined the low quality of service provided to the network subscriber: low data transfer rates, as well as significant and uncontrolled delays in the transmission of user information over the network. These features have virtually no effect on the functioning of traditional LAN applications, but sharply limit the use of the technology in applications sensitive to time delays — the transmission of voice information and SNA (Systems Network Architecture) traffic. It was possible to get rid of these shortcomings in the Frame Relay protocol.

The Frame Relay standard describes an interface for accessing networks with fast frame switching and provides functions for the first, partially second and third levels of the seven-level OIS system, including a small set of rules and procedures for organizing information exchange.

A distinctive feature of Frame Relay is the absence of error correction and data flow control mechanisms typical of X. 25. A frame received by an intermediate or end node with errors is discarded by the network, and error correction functions are assigned to higher-level protocols, such as TCP.

The mechanisms that ensure reliable data transmission in X.25 include acknowledgement and buffering of frames/packets that require confirmation, retransmission of incorrectly received frames, and some other algorithms. These procedures introduce additional delays, the magnitude of which depends on the probability of errors occurring at the data link layer of the data transmission link, as well as on the load on the buffer memory of intermediate nodes.

Refusal of these algorithms reduces network delays during user data transmission, and most importantly, allows them to be predicted. In addition, several service fields are excluded from the frame format. Thus, protocol redundancy is reduced, and, consequently, the efficiency of channel capacity use increases. This approach limits the permissible BER value to 10-7 and imposes stricter requirements on channel quality.

Subscriber access to the Frame Relay network is controlled by the User-to-Network Interface (UNI). Its main task is to describe the characteristics and features of multiplexing logical PVC (Permanent Virtual Connection) connections, as well as to control their status and configuration. Each such logical connection has its own unique number — DLCI (Data Link Connection Identifier).

As already noted, Frame Relay procedures do not provide for data flow control. Instead, the standard includes very simple congestion notification mechanisms that inform the user device that the network resources are almost exhausted. The Frame Relay standard defines the basic characteristics of each logical connection, according to which the system varies the rate at which the subscriber device transmits frames to the network.

The first such characteristic is the guaranteed data transfer rate CIR (Committed Information Rate) — the average speed of message delivery to the end device. The second characteristic — the guaranteed pulse volume of transmitted information Вc (Committed Burst Size) — determines the number of bits that can be transmitted to the network via PVC with a given guaranteed speed during time T. The last parameter — the additional pulse volume of transmitted information Be (Excess Burst Size) — specifies the number of bits that are transmitted to the network without a delivery guarantee. If network resources allow, then in addition to a message of size Вc, an additional data volume less than or equal to Be will be sent (at a speed exceeding CIR). All frames will be marked with the DE bit: for any network device, this bit means permission to drop a frame if the node is overloaded. The network will drop all information transmitted in excess of the volume equal to (Вc + Вe) (Fig. 3).

Fig. 3. Data flow control mechanism in Frame Relay technology

However, any frame discard entails its retransmission in accordance with the error correction procedures of higher-level protocols, which means that the network will still be overloaded. To inform the transmitter of the need to reduce the frame transmission rate, BECN (Backward Explicit Congestion Notification) service messages are created for the end devices of the virtual channel — they are transmitted to the source — and FECN (Forward Explicit Congestion Notification) — transmitted to the recipient. Receipt of such a message by a device means the need to reduce the data transmission rate in the network.

Advantages and disadvantages

Research shows that the use of TDM (Time Division Multiplexing) methods with rigid assignment of the physical channel capacity to logical connections allows the use of the digital data transmission channel resource by no more than 50°. Applications do not occupy the channel resource allocated to them permanently. On the other hand, when the resource is free, it cannot be given to another application.

Statistical multiplexing technologies, including Frame Relay, distribute the channel capacity between incoming data streams depending on the available network resources and predetermined characteristics of logical connections, allocating the required resource only for active applications. In addition, Frame Relay is well suited for transmitting pulsed traffic, as well as for combining data streams with different time characteristics, which distinguishes it from TDM (Fig. 4).

Fig. 4. Comparison of time characteristics of Frame Relay and TDM technologies

Since both X.25 and ATM also implement statistical multiplexing, the question arises: «What are the advantages of Frame Relay over these methods of organizing a transport system?»

Due to the peculiarities of its architecture, X.25 is unable to provide the time characteristics of logical connections required by applications that are sensitive to time delays. Frame Relay can combine various communication services, using the same network resource for transmitting both voice and data. In addition, due to the reduction in the amount of service information, this protocol uses the available channel bandwidth more efficiently.

ATM supports the same functions as Frame Relay, but this technology does not describe standards that provide for the transmission of information at speeds below 2 Mbit/s. If we take into account, firstly, the cost of a digital channel providing such bandwidth, and secondly, the price of the ATM equipment itself, it becomes obvious that the costs of implementing and further operating the same technical solution within the framework of these two technologies will differ by an order of magnitude. In addition, if we take into account the global trend in demand for communication services, we can see that for applications of 70% of primary users, speeds do not exceed 2.048 Mbit/s.

Frame Relay and ATM should be considered as a complement to each other. By organizing low-speed access to the ATM backbone network via the Frame Relay protocol, a communications operator can create flexible and productive information systems that provide the required quality of service.

The Frame Relay standard provides for the transmission of IP, IPX, SNA and X.25 protocol traffic, the organization of Ethernet/Token Ring network bridge connections, and the transmission of voice and video data.

When talking about multimedia applications, one cannot help but mention such a concept as «quality of service». This term refers to the provision of specified characteristics of logical connections for various applications.

For example, voice traffic is sensitive to delays, which means that buffering of frames containing speech information should be minimal. Unlike voice traffic, video data allows for significant channel delays and is insensitive to changes in its value, but it is very demanding on channel bandwidth, so it is necessary to ensure its redundancy.

Ensuring quality of service is a complex task, especially in conditions of transmission of heterogeneous traffic. It should be noted that, unlike ATM technology, Frame Relay standardization committees — in particular, the Frame Relay Forum — do not provide for QoS (Quality of Service) means for this protocol. In this matter, the Frame Relay Forum gives equipment manufacturers complete freedom of action.

Some developers believe that guaranteed service parameters should be provided by switching equipment of geographically distributed data transmission networks, but most manufacturers provide in their equipment functions that ensure prioritization and bandwidth reservation for different types of traffic, protocols and individual logical connections.

However, no method will be able to optimize the system's operation if the Frame Relay network is overloaded or the virtual channel parameters are incorrectly selected. Therefore, when purchasing the Frame Relay service from a telecom operator, it is necessary to know in advance which applications will distribute the virtual channel's bandwidth and what requirements each of these applications places on the connection parameters (including CIR and Bc).

The fact that, having connected once to the digital network resources, the subscriber will then be able to receive any number of new data transmission channels (within the capacity of the physical line and the allocated resource) by creating logical connections speaks in favor of the technology. In addition, the operator providing Frame Relay services ensures full control over the network status, overloads and failures in operation, guarantees rapid restoration and rerouting of logical connections in the event of a disruption in their operation.

No matter how broad the possibilities offered by any new technology, the user will not be able to appreciate it if it is too expensive to implement and operate. Simplicity, high level of standardization and flexibility of Frame Relay technology significantly reduce the costs of creating and operating information systems. By choosing Frame Relay, the user wins in the following:

• providing multiplexing of almost all types of traffic, Frame Relay allows you to abandon the operation of several parallel telecommunication systems (IP networks, X.25, switched telephone network) and move to a single digital network with integration of services;

• statistical multiplexing ensures efficient use of channel capacity by combining incoming streams whose total capacity may exceed the physical channel resource;

• reduction of protocol redundancy ensures minimal costs per bit of transmitted information;

• simple technology for organizing the transport environment reduces equipment costs and operating expenses, ensures higher reliability and system performance, and the simple procedure for reconfiguring network parameters does not require the involvement of highly qualified specialists.

Despite all its advantages, Frame Relay technology is not without some disadvantages. One of the most significant is the lack of a developed mechanism for establishing switched virtual circuits (SVC) in the equipment of leading manufacturers. SVC connection technology has a number of obvious advantages over traditional PVC connections:

• SVC-based solutions are more flexible and are implemented by the subscriber without the participation of the network operator;

• fewer logical connections are required within the overlay corporate network;

• provision is made for the transition to PVC operation in the case of large volumes of transmitted information and a flexible combination of these two methods;

• the user is able to establish a connection «on demand», that is, only when he needs to access network resources. In this way, he saves his money, and the provider saves the resources of his network.

Now the standard describing the switching of virtual connections via the Frame Relay network has already been adopted, and all that remains is to implement it in the equipment and begin providing services based on it.

Usually, the Frame Relay network operator charges for the following main parameters and service characteristics:

• access speed is the maximum throughput that can be obtained on the physical interface provided by the service provider;

• agreed information rate (CIR);

• traffic impulse characteristics (Вc and Вe);

• port fee determines the rental fee for the port provided by the provider for network access;

• terminal equipment rental. The equipment can be provided by the communications service operator and installed in the client's office — in this case, the operator resolves all issues related to the operation of the device. However, such a service is very rare, and, as a rule, the user uses his own terminal equipment;

• distance between access points in the Frame Relay network operator's network. If the logical connection ensures interaction between significantly remote objects, then the fee depends on the distance between access points;

• connection pricing. The cost of transmitting a unit of information is paid. Most often, the unit of measurement is 1 MB.

In order to correctly select the parameters of the communication channel and not overpay for excess bandwidth, it is necessary to know the characteristics and features of those applications whose traffic will be transmitted via the Frame Relay network. The most qualified assessment can be given by technical experts of the company-supplier of equipment for access to Frame Relay networks. However, the most general recommendations can be formulated as follows.

• For applications sensitive to delays, frame loss, and characterized by a constant traffic intensity over time, it is necessary to select a CIR equal to the total throughput within one logical connection and a minimum Be value.

• It is better to organize the transmission of pulsed traffic, such as local area networks, on the basis of virtual connections with a CIR equal to zero and a maximum Be.

Features of equipment selection

Modern Frame Relay equipment is produced by many well-known companies, including Northern Telecom, Motorola ISG, Cisco, Memotec. The set of functional capabilities of equipment from different manufacturers is almost the same. Therefore, the main emphasis in the chapter will be on the features of the implementation of specific technologies in FRAD devices.

Due to the demanding nature of voice traffic for the channel's time characteristics, the terminal equipment that provides voice transmission must perform some additional functions.

During data transmission to the network, a situation may arise when a large data frame appears before a frame with voice information in the device buffer. Processing a long frame will require a lot of time, therefore, a delay will occur in transmitting the voice information frame. If such a situation is repeated and the length of large frames varies, extraneous pauses and freezes will appear in the voice communication session. To avoid such an undesirable effect, a traffic segmentation mechanism is used. This algorithm has one drawback: in the single-frame format, the ratio of service information to useful information increases with the inclusion of segmentation, which means that the efficiency of using the data transmission channel bandwidth decreases. Therefore, the end device must recognize speech information in the data stream, turning on segmentation if necessary and turning it off in the absence of speech.

In addition, to ensure minimal delay in frames with voice information, FRAD devices must provide the ability to prioritize different types of traffic.

Another important feature of the operation of access equipment in the Frame Relay network is the ability to multiplex voice connections and data traffic within a single PVC (Permanent Virtual Circuit). Some manufacturers of FRAD devices separate the transmission of voice messages and data traffic over different logical connections, each of which is assigned a separate DLCI. This significantly increases the cost of the system, since the operator charges a fee for each logical connection. Multiplexing data within a single PVC provides an additional economic effect. In addition, modern equipment that performs FRAD functions has built-in voice compression mechanisms that ensure even more efficient use of the bandwidth of data transmission channels. To transmit voice, the FRAD device must have a DSI (Digital Speech Interpolation) pause suppression function. Its operating algorithm determines the presence of speech activity, and then voice information frames are transmitted to the network. In the absence of speech, the bandwidth resource can be used for another voice message or data transmission.

When discussing technologies for transmitting heterogeneous traffic via Frame Relay networks, we cannot help but touch upon such an issue as organizing the interaction of local networks. The simplest method — organizing bridge connections — is not the best. This is explained by the fact that the broadcast nature of the traffic, as well as the service information of LAN protocols, significantly reduce the available resources of the data transmission channel. Therefore, the most acceptable solution to this problem is organizing the interaction of LANs at the network level. This approach allows us to sharply reduce the number of broadcast messages, as well as get rid of the service information of the channel layer protocols.

Forecasts and Prospects Despite the fact that standards for voice transmission over Frame Relay networks have existed for a long time, not all equipment manufacturers support them in their products, implementing proprietary algorithms for processing voice information1. Therefore, an important step in the development of Frame Relay will be the creation of unified mechanisms for interaction between network devices during voice transmission that are independent of the equipment model. According to forecasts by Vertical System Group, voice information transmission (VoFR — Voice over Frame Relay) will continue to be one of the fastest growing services in the Frame Relay market (Fig. 5).

Fig. 5. Fig. 5. Projected growth of VoFR services

In general, based on the global experience of Frame Relay technology development, we can note its smooth convergence with ATM-based solutions. Such integration is due to the fact that corporate networks operating on the basis of virtual channels of digital networks have a star topology. With such a configuration, remote offices do not need high bandwidth to meet the needs of all their applications. At the same time, all logical connections are concentrated in the central node of the network, which occupy most of the resources of the user's interaction channel with the network. The most effective in this case is to use the Frame Relay protocol for low-speed connections and ATM to concentrate traffic in the central node. Telecom operators solve similar problems when providing clients with access to various network resources.

The main problem with the process of merging technologies is that Frame Relay has not developed standards similar to ATM standards to ensure «quality of service». A possible way out of this situation is to combine the QoS mechanisms of Frame Relay technology, implemented in the equipment of some manufacturers, and technical solutions based on ATM equipment. This approach to building data transmission networks allows for flexible and scalable solutions, supporting a full range of possible applications and communication services.