Ethernet Applications and How Microwave Radios Can Play a Part
By Dirk Heitkamp and Alan Hammock
Introduction
If you look anywhere in the communications industry these days, you will invariably notice that Ethernet communication products and applications are becoming more entrenched as the new standard for data transport. What is the reason? Carriers are cutting costs by eliminating the need for overlay networks that require different equipment for different service types such as Circuit-Switched, Frame Relay, and remote access services. The primary goal for carriers is to accommodate all of these services on a single network and Ethernet IP (Internet Protocol) appears to be the clear winner. The main advantage with this technology is that carriers will no longer incur the added costs associated with overlay networks such as:
* Separate maintenance staff.
* Additional capital expenditures.
* Training for maintenance staff.
* Separate billing structures.
* Separate customer services staff.
* Separate installation staff.
* Separate inventory.
In general, Ethernet networks cost less to build due to lower cost switches and they handle data transfer more efficiently then the traditional circuit-switched, time-division-multiplexed (TDM) alternatives. Ethernet is more efficient because it only uses bandwidth when data is transmitted. TDM systems use full bandwidth whether data is transmitted or not.
The scope of this of this paper is to describe the technologies and applications associated with Ethernet IP and to show how Ethernet microwave radio can be used to build or supplement new network implementations faster and more economically than most other alternatives. Specifically, the applications and technologies discussed are Ethernet IP, Local Area Networks (LAN), Virtual LAN (VLAN), Virtual Private Networks (VPN), and IP Telephony or Voice over IP (VoIP).
Introduction to LANs
LANs (Local Area Networks) were created out of necessity to connect workstations in an office setting, allowing workers to communicate via internal email, transmit files, and share software and printers. In addition to the impact that personal computers have made in the workplace, LANs have also greatly contributed to the surge in productivity that started in the 90s.
A LAN can simply be defined as several users interconnected by a common connecting medium. Different media have been used during the evolution of LANs, but most have now settled on twisted pair cable. Each user terminal is called a node and has unique digital identity called an address. Data transmitted from any node travels the connecting path to all of the other nodes on the network in short bursts called packets. In addition to the actual data, each packet includes the address of the sender and the intended recipient. As the data travels across the network, nodes whose address does not match the recipient address on the packet simply ignore the data.
The most popular architecture used by modern LANs is called Ethernet. Ethernet was developed by Xerox in partnership with Digital Equipment Corporation (DEC) and Intel in 1976. Later, this specification served as the basis for the IEEE 802.3 (Institute of Electrical and Electronics Engineers) standard, which specifies the physical (hardware) and lower software layers of the Open Systems Interconnection (OSI) model. Ethernet uses the CSMA/CD (Carrier Sense Multiple Access / Collision Detection) system to handle data traffic on a network. This is basically a protocol that checks to see if the network is available before a device sends its data packets onto the network. If the network is not busy, the data is sent. If two devices send information at the same time, a collision is detected and the process starts over again.
One problem encountered with traditional LANs is that they cannot grow with companies as they expanded into other buildings not geographically close together. To correct this limitation, MANs (Metropolitan Area Network) and WANs (Wide Area Network) were then defined to extend the reach of corporate LANs. MANs usually refer to data networks across a city or town. WANs tend to expand to anywhere in the world and are often connected through the public telephone networks. The Internet is considered the largest WAN in existence.
When two or more LANs are connected to form a MAN or WAN, it is usually facilitated through a network of devices called routers or switches. These devices sort through the data traffic and selectively send data packets only where they need to go based on the IP address in the packet. Packets intended for remote LANs are allowed through the router while all others are not. This in effect optimizes the network traffic by limiting data to LANs of similar interest and subsequently reduces overall network data congestion.
VLAN Overview
A VLAN is a technique to segment different workgroups connected to the same physical medium. The functional implementation of a VLAN is configured through software rather than hardware, which makes them extremely flexible. A VLAN is a software grouping of PCs, servers, and other network resources that behave as if they were physically connected to a single LAN segment – with packets for one workgroup not visible by another workgroup. Several separate, mutually exclusive VLANs can share the same physical infrastructure while maintaining anonymity and security from one another. For example, all marketing personnel may be spread throughout a building. Yet if they are all assigned to a single VLAN, they can share resources and bandwidth as if they were connected to the same LAN segment. The resources of other departments assigned to a separate VLAN can be invisible to the marketing VLAN workgroup. VLANs are highly configurable at the IT Manager’s discretion, e.g. access privileges, and bandwidth allocations.
This virtual grouping of network nodes helps to free IT managers from the restrictions of their existing network design and cabling infrastructure. It provides a fundamental improvement in the way LANs are designed and managed. Since VLANs are software-based, they allow the network structure to quickly and easily adapt to organizational changes, reassignment of resources, and accommodation of multiple organizations in the same physical proximity.
VLANs also help improve network efficiency by segmenting nodes through selective dissemination of data traffic. This reduces the burden of irrelevant traffic on other network devices. If the node is not on the VLAN, the data is ignored. Security can also be improved since all packets that travel between VLANs may also travel through routers. In those instances, router-based security measures can be implemented to restrict access as needed.
VLAN Implementation Techniques
Prerequisite to understanding this section is some knowledge of packet-based protocols. Generally, there are three different techniques used to implement VLANs between different workgroups – port-based, MAC address-based, and layer 3 or protocol-based.
Port-Based VLANs – the administrator assigns each port of a switch to individual LANs to identify packets. The switch then determines the VLAN membership of each packet by noting the port on which it arrives. When a user physically moves to a new location and subsequently to a different port of the switch, the administrator can simply reassign the new port to the user’s old VLAN. The network change is then completely transparent to the user with no wiring changes by the IT manager.
The primary limitation to port-based VLAN workgroup assignments is when the user is moved to a location where a new port in the switch is not available or is not of the same VLAN. Wiring changes are required in this situation.
MAC (Media Access Control) Address-Based VLANs – uses the embedded hardware address of each node to identify packets. Every network device connected to a LAN has a unique MAC address assigned by its manufacturer. Membership of a MAC address-based VLAN is determined by its source or destination MAC address. Each switch must maintain a table of MAC addresses and their corresponding VLAN memberships. A key advantage of this method is that the switch doesn’t need to be reconfigured when users move to a different port location. Assigning and maintaining accurate VLAN membership tables is a time-consuming task. The primary limitation to MAC address-based VLANs is the difficulty of assigning a single MAC address to multiple VLAN workgroups, making it difficult to share server resources between more then one VLAN.
Protocol-Based VLANs – use the IP or Layer 3 addresses to route/switch packets between different workgroups. This is the most flexible method and provides the most logical grouping of users. For example an IP subnet can be assigned to their own VLAN. Since there is no manual intervention required when users are relocated, this method has the greatest flexibility for the IT manager.
When a packet travels between switches, the implementation method used in both port-based and protocol-based VLANs is explicit. This simply means an identification tag is added to the packet indicate its VLAN membership. IEEE 802.1q VLAN specification defines the interoperability of VLANs between switches and NICs (Network Interface Card).
VPN
The VPN or Virtual Private Network is a result of VLAN technology as described above. VPN is a data service where several “data communities” share a common transmission medium but the data belonging to each community is available only to members of that community. The transmission media may either be a privately owned and operated network or, increasingly, the public Internet. The Virtual Private Network provides fast, reliable and secure world-wide communications for companies at lower costs than previous technologies.
Because they supply network connectivity over long distances, VPNs are a form of WAN. The Internet Engineering Task Force (IETF) defines VPNs as “the emulation of a private wide area network using public facilities, including the public Internet or private IP backbone.” Companies have sent private data over public networks using frame relay or ATM for years, but using the Internet as the data network is new. From a user’s point of view, VPNs act like private lines, although data are actually transported over shared networks. There are several techniques used to ensure security such as encryption and encapsulation.
VPNs are typically used in intranets and extranets. Intranets are private VLANs and WANs that enable geographically dispersed workers to access corporate databases without dedicated leased lines. Extranets provide secure connection between an organization and its customers, suppliers, or business partners. VPNs also serve the rapidly increasing market for remote access, allowing mobile workers to connect easily with corporate servers over any Internet connection. In addition to cost savings, VPNs are available worldwide.
IP-VPNs provide significant savings compared with frame relay network because fewer connections are needed. Companies connecting their corporate location via frame relay have an IP connection for Internet traffic and a separate frame relay connection for WAN traffic. If the WAN traffic can go over the IP-VPN along with the Internet traffic, the frame relay connection can be eliminated. IP-VPNs are also less expensive than private lines, and they improve on traditional VPNs by giving companies more bandwidth, more flexibility and easier connections with remote offices.
MPLS/Carriers
VPNs are not limited to the business services sector. Carriers are beginning to use a technology hyped for years called MPLS (Multi-Protocol Label Switching) to supply the growing demand for services that “converge” voice, video, and data onto a single network. This move is enhanced even more with the emergence of VoIP (discussed later), which reduces the importance of traditional voice networks.
MPLS alleviates many of the problems that arise in shifting applications traditionally run on circuit-switched network to IP-based networks. Because MPLS can handle any type of traffic, carriers can use it in their core infrastructure to converge traffic from all of its networks onto a single network, saving money in operational costs.
The main advantage of label switching is that it allows routers to make forwarding decisions based on the contents of a simple label, rather than by performing a complex lookup based on the destination IP address.
In an MPLS network, incoming packets are assigned a “label” by a router. Packets are forwarded along a label switch path where each router makes forwarding decisions based solely on the contents of the label. At each hop, the label is stripped off and a new one is added that tells the next router how to forward the packet.
In many ways, the label switched paths of MPLS are no different than circuit-switched paths in traditional telecommunication technologies such as ATM (Asynchronous Transfer Mode) or Frame Relay networks. The main difference is that MPLS paths are not dependent on a particular transport technology. This means that MPLS can be used with any transport technology including ATM, Frame Relay or Ethernet. Thus, one of the true promises of MPLS is the ability to create end-to-end circuits across any type of transport medium, eliminating the need for overlay networks.
IP Telephony/VoIP
Another technology that is quickly becoming more popular is Voice over IP (VoIP). VoIP is a technology that offers the potential for considerable cost savings to service providers and businesses of all types, and is the main enabler to convergence as described in the last section.
As the name implies, VoIP uses IP packets to transport voice over an IP data network. Traditionally, voice was transported exclusively over the public switched telephone network (PSTN). The main disadvantage of this transport technology (circuit-switched) is the inefficient use of bandwidth. Each voice channel in a circuit-switched TDM network ties up 64 kbps of bandwidth as long as the connection is made, even during times of no actual conversation. Bandwidth efficiency in an IP network is considerably better.
In VoIP, each conversation is converted from analog to digital by a codec (coder-decoder). The digital bits are then loaded into IP packets ready for transmission. During those moments where there is no conversation no packets are created and no bandwidth is required. Effectively, where TDM always required 64 kbps for each conversation, VoIP only requires 10 to 15 kbps of data. This efficiency does come at a price. Voice packets cannot tolerate the randomness often associated with connectionless technologies such as IP. Voice is particularly affected by packet delay and jitter in the network. Since many earlier IP networks were only concerned with data transport delay and jitter were not tightly controlled.
It is obvious from this discussion that if delay (or packet latency) and jitter play a critical role in voice quality, IP networks that will include VoIP traffic must use special delay management measures to mitigate these affects. Today’s IP networks can be designed to offer PSTN quality service. This requires the network to recognize a voice packet and apply differentiated treatment to voice versus data packets.
Voice and data differ in their tolerance for delay, jitter, packet loss, and echo. To better understand the issues of transporting voice, we need to discuss these in more detail.
* Delay – The amount of time elapsed between when a packet is transmitted from the source and when it arrives at its destination. When one-way delay is more then 250 ms, the transmissions in opposite directions tend to overlap and make conversations difficult.
* Jitter – Jitter measures variation in the difference between inter-packet arrival times and inter-packet departure times. For received voice to be of acceptable quality, variation in jitter as well as the jitter itself must assume low values. To remove unacceptable jitter, packets must be held in buffers for a sufficiently long time to smooth out the variations in inter-packet times. Thus the goal of reducing jitter conflicts to some extent with that of reducing delay.
* Packet Loss – Occasionally, random packets may be “lost” in the network and never arrive at the destination. Packet loss is one area where the requirements for voice traffic are less stringent than for data. Techniques of adapting to packet loss include replacing a lost packet with the last packet received and sending redundant packets. These options are not available to data transport where loss of a single packet is potentially disruptive.
* Echo – Echo refers to the phenomenon of the user hearing his or her own voice and is caused by reflections from the far end. If the round trip delay through the network is greater than about 50 ms, echo becomes a problem. This is not a serious concern in circuit switched networks where delays are small, but becomes significant in packet switched IP networks, and requires the use of echo cancellation techniques.
Fortunately, mechanisms are available in the form of quality of service (QoS) techniques to mitigate these effects. These solutions are implemented by configuring network components such as routers and switches. IP QoS partitions network traffic into classes of service by utilizing the first three bits in the eight-bit Type-of-Service field within the IP protocol header. These three bits are called the IP Precedence field as defined by IEEE 802.1p. Each class of service can be assigned a ranking or priority. Using this method, a switch or router can reorder data packets to allow fastest delivery of priority packets.
MPLS is another tool to help ensure QoS in mission critical networks. MPLS uses special routers and switches to insert a label between layer 2 and layer 3 protocols that identify packets requiring special handling. The labels not only contain routing information, but also contain priority information.
In a typical scenario, a service provider defines three or four classes of service and identifies one as appropriate for voice traffic, equipping it with suitable delay and jitter characteristics. These characteristics are ensured by configuring the available queuing, traffic shaping, and policing mechanisms on the routers.
IP is an OSI layer 3 protocol. Voice transport typically requires two additional layers. In particular, it utilizes the Real Time Protocol (RTP) on top of the User Datagram Protocol (UDP), which, in turn, sits on top of IP.
Real Time Protocol (RTP) encapsulates voice or video traffic in its data field and utilizes the connectionless mechanism supplied by UDP to transport voice or video traffic end-to-end. Real Time Control Protocol (RTCP) allows the communicating source and destination to monitor and control the quality of the communication between them.
One further item of note is that all QoS methods have a somewhat lesser role to play when there is no congestion in the network, although they can improve bandwidth utilization and throughput at any level of traffic. But it is at the points of congestion where bandwidth guarantees and priority to certain packets over others become critical.
Microwave Transport
What does all this have to do with Microwave Radio? Microwave radio systems are used as an alternative transport technology for voice, data, and video applications. It can be used to create complete networks or just to carry out the last mile connections of larger wire-line or fiber networks. All of the applications stated above can be transported using microwave; however, IP data transport is a relatively new access type for these products.
Microwave systems are used by businesses of all types for many different applications. Some examples include:
* Mobile Carriers use microwave for base-station backhaul of voice traffic for route to local PSTN.
* Telcos, like the former Bell companies, use microwave for redundant backup on critical paths, accessing remote areas to supply voice and data services, provide dedicated access to corporate users with high traffic volume, or for creating and completing VPNs for corporate users.
* ISPs are using microwave to create their own networks to reduce their dependence on telcos to gain access to their customers.
* Long distance companies also use microwave to bypass the local telco to eliminate access charges.
Obviously the potential applications for microwave are as vast as the many different types of communication services.
Some of the advantages that Point-to-Point (PTP) microwave systems offer include:
* Faster network rollout and implementation (no trenches to dig)
* Lower cost of ownership over time or return on investment (ROI)
* Increased network design flexibility
* Full management & control over operator owned network
* Relatively low initial investment
* Low operating costs
* No long outages due to so-called “back-hoe fade” (accidentally digging up cables)
* Better resilience to natural disasters.
From the list above you can see microwave radio implementations have many advantages over leased lines from the public phone networks. Here is a typical example of a school system’s move from leased lines to microwave radio.
Case Study:
A public school system has six campuses and an administration building in its region. Each school has its own PBX with 25 to 100 telephones at each campus. Initially, leased lines are used to connect the PBXs at each campus to the local telephone company central office.
Eventually, Ethernet LANs were added at each campus and then connected to each other to form a WAN. The WAN is also connected to the local Internet Service Provider (ISP) by leasing DS1 or DS3 lines. The WAN is further upgraded to add video conferencing for inter-campus sessions and “distance learning.” To maintain adequate connectivity the network requires even more DS1 and DS3 lines.
Obviously, the network for this school system has grown considerably past the need for just local phone connectivity. Now if microwave radio is used to replace all the leased lines currently in use, what is the effect?
Eliminate Leased Line Costs
Rather than leasing DS1 and DS3 lines for each campus, a one time purchase of microwave radios can replace those reoccurring costs with a payback from one to two years. Also, since microwave can now carry DS1 and Ethernet simultaneously, one radio link is all that’s needed for the two different network architectures.
Abundant Bandwidth
The latest technologies in microwave can provide for large amounts of data transfer. It is unlikely that the data capacity for these new radios will ever be exceeded. This is especially true in Ethernet applications where data is “bursty” and does not require full-time network access.
Bandwidth on Demand
Since all campuses are networked together, only one connection is required for Internet access to the local Internet Service Provider. If an Ethernet connection is used instead of the traditional DS1 lines, most ISPs can provide high speed connectivity on an as-needed basis for such things as videoconferencing or other applications that do not require full time access. Unlike a leased DS1, with Ethernet you pay only for the time and bandwidth you need.
Path Protection
Using leased DS1 lines from each campus to the telephone company makes each campus vulnerable to network outages due to hardware failures. With microwave radio, a ring network can be established where traffic from each campus is relayed from campus to campus in a complete circle. If the connection between any two campuses should be lost, traffic can automatically be rerouted around the ring in the opposite direction to reestablish the connection.
Path Reliability
Most telephone companies offer 99.9% reliability on each connection – an average of about 1 hour unavailability per year. A properly designed microwave path typically offers 99.999% reliability – 5 minutes per year unavailability. In a ring configuration even in the rare event of equipment failure traffic will be rerouted to bypass the damaged equipment.
Although this hypothetical example involved a local school system, the growth path and implementation could easily apply to large business organizations whose buildings are spread across a relatively large geographic area.
Summary
Microwave radio is a flexible and cost-effective alternative for transmission of voice, data, and video services in all parts of a fixed or wireless mobile network, including applications for the backhaul or direct access services. With the scale and flexibility of today’s new radio technology, implementing a microwave network is more economical and easier then ever. A typical microwave link can be installed in just a few days. Enhancements in wireless technology have also allowed higher order modulation schemes to be implemented that offer spectral efficiencies not possible in higher frequency radios just a few years ago. These efficiencies permit less bandwidth for the same capacity or higher capacities within the same bandwidth compared to older models. All this gives network designers and operators many more choices to build and maintain their networks.
Microwave Networks Incorporated provides fixed wireless products for next generation “converged” networks.
MNI’s product family is focused on the challenges facing all network operators as they consider their options to reduce network inefficiencies and increase profitability. Our forward thinking product platforms allow for the greatest flexibility in network design as older networks are converted to handle the new Ethernet IP architecture that so many operators have now adopted for all network traffic.
Our newest radio product called Proteus AMT is a good example. This product provides multiple Line Interface Modules (LIMs) that give user access to a variety of Ethernet plus T1/E1 configurations and can support data speeds up to 100 Mbps.
Proteus AMT VLAN Features
The Ethernet LIM is unique in that it incorporates many of the same functions as a Router or a Switch. This LIM includes four configurable IP ports with standard RJ45 connectors for 100BaseT Ethernet input/output. The ports can be configured as access ports on a single LAN or can be individually configured into VLANs based on any of several criteria. In the VLAN mode the ports can be set to route incoming data based on the port which the data arrives, it can route and prioritize data packets based on information in the IP header, or can route based on the VLAN tags.
To further explain these functions in detail, the following is a list of functions and their configurations supported by the Proteus AMT Ethernet LIM:
Port 1 to Port 4 – The 100BaseT LIM has four customer access ports. If VLAN is disabled the four ports act as 4 switched ports on a single 100 BaseT LAN. With VLAN enabled each of the four ports is independently configurable. Two 100BaseT LIMs can be installed together in a radio, providing up to 8 ports.
Default VID – In the VLAN mode each arriving packet is assigned a Default VLAN ID (VID) based on which of the four ports the packet enters. In the port-to-port VLAN configuration data arriving on a specific port at one end of the radio hop is transmitted only to a designated port at the far end of the hop based on the Default VID. Normally the radio will be configured for near-end port 1 to transmit to far-end port 1, port 2 to port 2, and so on, but the system can be configured differently if desired. In the port-to-port configuration any VLAN tags already attached to the arriving packets will be ignored, although the packet will be transported across the hop with the tags intact.
Traffic Class (TC) – In the VLAN mode, incoming (ingress) packets on each port can be divided into four Traffic Classes: TC0 through TC3. The Proteus AMT treats TC3 with the highest priority and TC0 the lowest. The network administrator maps incoming packets into the different traffic classes based on the Type of Service bits or the 802.1p priority bits (explained below). On each port the amount of throughput allocated to each Traffic Class can be designated. For example on Port 1, TC0 may be limited to 1 Mbps, TC1 limited to 4 Mbps, TC2 limited to 4 Mbps, and TC3 limited to 8 Mbps. The total required throughput for Port 1 is therefore 1 + 4 + 4 + 8 = 17 Mbps. This means that in a radio configured for 100Mbps total throughput, 83 Mbps remain available for the other 3 ports.
Traffic Class Weighting – For transmission across the radio hop (egress) there are two ways the LIM can operate. With Traffic Class Weighting disabled, the Traffic Class 3 packets (from all 4 ports) will always be sent ahead of other Traffic Classes. After all TC3 packets are sent, TC2 packets will be forwarded, then TC1 and finally TC0. If a large percentage of network data consists of TC3 packets, TC0 packets may transverse the hop very slowly. Enabling TC Weighting prioritizes the packets with an 8-4-2-1 priority: for every eight TC3 packets sent, four TC2 packets, two TC1 packets and one TC0 packet are sent. Weighting guarantees some level of egress for lower weighted packets even in the face of a high traffic congestion of higher priority packets.
Traffic Class Order – The network administrator decides which Traffic Class each packet is mapped to based on three criteria: Default VID, Type of Service, or 802.1Q Priority.
Default VID – As defined earlier, with the Default VID method packets are mapped to a specific Traffic Class based upon the LIM port they arrive. This arrangement would be useful if, for instance, you have a VoIP server connected on Port 1 and an e-mail server on port 2. You would normally want the VoIP to take priority so you would map Port 1 to TC3 (highest priority) and Port 2 to a lower TC.
Type of Service (TOS) – The Internet Protocol (IPv4, IPv6) defines 8-bits in the packet header for specifying Type of Service (also referred to IP Precedence bits). The LAN/WAN network administrator determines how these bits are used in a specific network. For example TOS 0 may be defined for Voice over IP packets. TOS 64 may be defined for HTML frames. E-mail data may receive TOS 128, and so on. In the Proteus AMT a range of TOS is mapped into each of the 4 Traffic Classes. For example TOS range 0 – 31 may be mapped into Traffic Class 3; TOS 32 – 63 mapped into TC2, 64 – 127 mapped into TC1, and 128 to 255 mapped into TC0. In this manner a packet arriving with a TOS of 0 would be mapped to TC3, the highest priority traffic class.
802.1Q Priority – For packets with external VLAN tags (tags added externally by equipment other than the Proteus AMT), the tag includes 3 bits used to identify the packet’s priority from 0 to 7. These bits are sometimes called the Differentiated Service or DiffServ bits. In the Proteus AMT the 8 possible priorities are mapped into the 4 Traffic Classes. For instance, Priority 0 and 1 may be mapped into TC0, priorities 2, 3, and 4 into TC1, priorities 5 and 6 into TC 2 and priority 8 into TC3.
Ingress Rate – Each of the four ports can independently be set to allocate bandwidth based on the Traffic Class. For example, on LIM Port 1 TC0 could be limited to 256 kbps, TC1 to 1 Mbps, TC 2 to 2 Mbps and TC 4 to 4 Mbps, while on Port 2 TC0 might be limited to 1 Mbps, TC 2 to 2 Mbps, etc. The total cumulative ingress of all Traffic Classes on all 4 ports is limited to the maximum Ethernet throughput of the radio.
Conclusion
MNI understands the challenges that operators face when they begin to migrate their established network technologies. This is why all of our product platforms are designed with the greatest configuration flexibility and performance that enable the most efficient network growth as demand increases. And with our focus on Ethernet IP radio products, we are well positioned for the next generation.
For more information contact 281-263-6500 or visit www.microwavenetworks.com.