Unlocking the mystery of MPOA - Multi-protocol over ATM network architecture - includes Network Management Directory of companies listing contacts, phone numbers, matrices - Special Focus: Network Management - Technology Information - Cover Story
The arrival of Multi-Protocol Over ATM (MPOA) on the networking scene has been largely unheralded, considering its numerous benefits. In part, the problem rests with the name.
Mentioning it can prompt a response of, “Isn’t that what ATM’s supposed to do?” After all, most people expect ATM to be able to transport multiple protocols such as IP and IPX.
Also, many networks already transport a variety of protocols via another ATM technology called LAN emulation (LANE). So why is MPOA such a breakthrough?
Like any other technology, MPOA is designed to solve a problem. In this instance, it’s the problems created by today’s router-based networks, which are being swamped by the increasing wave of bandwidth-hungry applications demanded by users.
Related Results
Consider the evolution of a typical network, which began with a collection of users on a common, shared LAN — for example, an Ethernet segment. As the enterprise grew, additional segments were added and connected with bridges, then routers.
The number of and reliance on routers increased steadily over time as more and more users and applications were added to the network. As bandwidth demands outstripped the capabilities of Ethernet, a new backbone technology such as FDDI was implemented to solve the problem and maintain service guarantees.
Unfortunately, the reward for good network performance is demand for even greater performance — to support Web-browsing, whiteboarding, scheduling, multi-disciplinary teams, corporate intranets and, in the near future, multimedia that would amaze even the most jaded 13-year-old cybergame prodigy.
The-bottom line: raw bandwidth is no longer enough in a world where the network has evolved from useful business tool to strategic necessity. Today, network administrators must provide the bandwidth, along with several classes of traffic contracts with tight service guarantees.
The result is increasing pressure on the routers at the core of the typical network, accompanied by delays and painfully slow service that threaten the ability to meet those guarantees.
The immediate reaction is to add more routers — but it’s not necessarily the correct and certainly not the most cost-effective solution. Routers are expensive — a number of user studies peg the cost of router ports, in terms of price/performance, at five to 18 times the cost of ATM switch ports.
The price difference stems from the fact that routers rely on software and work in a connectionless paradigm. Most data passing through a router must be analyzed to determine where it will be directed and to check items such as filter lists — tasks typically performed by microprocessors running software. This analysis is neither cheap nor scalable, but it’s a necessity for every data packet transmitted.
Performance and cost are just two of the reasons why an increasing number of network architects are opting for ATM backbones. Since the early 1990s, ATM has demonstrated its value in thousands of mission-critical networks as an optimal solution in terms of price/performance, tighter quality of service (QoS), scalability, and inherent support for hybrid networks comprising data, voice, and video traffic.
Despite the advantages of ATM, two concerns continue to affect network architects’ decisions about adopting the technology: the need to provide ongoing support for non-ATM connected users, and the pressure to preserve existing network hardware and software investments.
Supporting non-ATM users is accomplished with the implementation of LAN emulation (LANE) over ATM, which allows an ATM network to emulate a physical LAN segment without the physical constraints, such as distance or maximum bandwidth, of a real LAN segment.
LANE is built entirely around Layer 2 of the protocol stack and deals in MAC addresses, making it completely transparent to Layer 3 protocols such as IP, just like the real LAN segments it emulates. In a typical LANE network, user devices are split across several of these emulated LANs (ELANs).
LANE enables network architects to build networks with an ATM backbone that supports Ethernet, token-ring, or FDDI clients connected via the legacy LAN to ATM switches, while servers and other high-bandwidth devices can be connected directly to the ATM backbone.
Traffic flows between users in an ATM virtual circuit, in hardware. The software-based problems of a router-based network disappear because 1) ATM using LANE adds a connection-oriented scheme that eliminates the need to analyze data packet-by-packet in every node, and 2) traffic forwarding is accomplished in hardware.
End of story? Not quite. Until recently, users in different ELANs still required a router that had been upgraded with an ATM interface to communicate, exposing the ATM network to the same old router problems.
Hence the need for MPOA. MPOA provides the AtM-based, connection-oriented path required for realizing the benefits of AtM — for example, QoS.
VIRTUAL ROUTERS
MPOA logically divides the operations of a traditional router in two and allows each function — route calculation and packet forwarding — to reside on different hardware platforms. With MPOA, these functions can be distributed throughout the network, forming virtual routers and providing important benefits: