CIS307: Introduction to Network Architectures

[Access Networks, Autonomous Systems]

Networks encompass a variety of technologies, are created and maintained by large number of ever changing industries, and must satisfy a significant number of often conflicting requirements. As Comer indicates "No single networking technology is best for all needs". A fundamental aim is to support Universal Service, that is, to allow any two computer to communicate, no matter the technologies they use and the specific networks they are directly connected to, as long as there exists a communication path between them. Internetworking is the ability to communicate across networks, with connection between networks provided at the network layer [see below] by routers (a router helps choose a "good" route from source to destination) or, at the data link layer, by bridges and by switches. An internet is a collection of internetworked networks. The Internet is the name for the global, public internet connecting most networks and using the TCP/IP family of protocols. All the entities connected in a network are called nodes. The computers attached to the network are called end systems, or hosts, or Data Terminal Equipment (DTE). The intermediate nodes of the network are called Intermediate Systems (ISs), or Interface Message Processors (IMPs), or Gateways, or Switches (Bridges and Routers).

Some significant characteristics of networks are:

Extent: The physical space covered by the network: Local Area Network (LAN), covering hundred of yards up to a few miles; Metropolitan Area Networks (MAN), covering up to a few tens of miles; and Wide Area Networks (WAN), possibly covering our planet.
More recently a new type of network has made its appearance: Personal Area Networks (PAN). It is a (usually) wireless network around a person (a few meters) involving devices like a computer, telephone, PDA, connecting them to each other and, possibly, to larger networks. Bluetooth is one of the technologies used in these networks. A related concept is Ad-Hoc Network, that is, a set of computers with wireless cards, that broadcast in their neighborhood, identify each other, and establish a network among themselves.
Ownership: Who owns the network: Public, owned by a state licensed (and state regulated)entity that will sell services to users; and Private, owned fully by a company for its private use (it may use cables leased by a service provider). A hybrid of the two is represented by Virtual Private Networks, where private sites are connected with each other using special boxes and protocols and public lines. The security of the data transmitted through the public lines can be guaranteed with encryption and routing restrictions. Also the quality of service, things like minimum data rate and maximum latency , may be guaranteed in a virtual private network.
Service: The way information is exchanged between users: connection-oriented and connectionless. In the former case a connection is established between the communicating agents and on that connection messages are exchanged (the connection can be persistent, i.e. fixed usually in hardware, and switched, i.e. that can be set/reset rapidly with commands.). In the latter instead each message is treated as a free standing entity (datagram). When a connection is established, the connection receives an id or address that is used for routing the packets involved in the connection. Related concepts at the communication level are circuit switched and packet switched. In the former the circuit is set at the wire level [the wire can be multiplexed among circuits using FDM or TDM] and communication between interlocutors always follows the same path. In the latter, individual packets follow their own independent routes, or, if a virtual circuit is set for the duration of the connection, they follow the same virtual circuit route.
Quality of Service(QoS): Here we refer to things like delays, reliability, jitter, and throughput. Delays can be due to various causes, we have propagation delay (time for transfers "on the wire"), switching delay (time to move across switches, bridges, routers), queueing delay (time waiting queued at a node waiting for a transmission channel to become available), access delay (time to wait to insert a message, as in Ethernet, onto a shared medium). Jitter, that is the variation in the duration of delays, such variation is or not significant depending if we are sending data in a backup (it is not), or in a real time-application (it is), or we transmit voice or video (it is). Throughput, that is the data rate at which information is actually transmitted. Reliability: that is, the availability of service and the infrequency/absence of errors. Reliability, defined as Guaranteed Uptime [thanks to Colon for bringing it up], is measured either as percent of the time when a service is up, or in days per year when the service is available. For example we can say that the guaranteed uptime is 99%, or more usually, 361.35/365 days. Notice that at 99% guaranteed uptime the service will not be available for over three days, much longer than people like. A more usual guaranteed uptime is 99.9% or 364.64/365 days.

We will see that protocols and standards play a major role in networking. A number of issues should be kept in mind when we analyze a protocol and its implementation. Among them:

Framing: How do we recognize the beginning and end of a message (or packet, or frame, or ..);
Flow Control: How do we make sure we don't send too much information to the receiver so that it cannot handle it and has to "drop it on the floor"; it is normally managed by the receiver that informs the sender about how much data it is ready to accept;
Congestion Control: How do we make sure that we don't have too many messages trying to get into a communication channel. Thus causing delays and perhaps lost packets; it is normally managed by the sender, and possibly by the routers;
Multiplexing: How a single communication channel can be shared for more than one conversation;
Addressing: How do interlocutors address each other [host interfaces are known by their IP address (a 32 bit integer); unfortunately in ethernet the address understood by the hardware is the Ethernet address (a 48 bit integer) of the connector; thus there is need of a way to translate from IP to Ethernet addresses. In general, this requires an address resolution protocol (ARP is such a protocol).
Error Detection and Correction: what we do to detect if an error has occurred in transmission and how to correct it (recover from it).
Fragmentation: One message at a layer may have to be fragmented over multiple messages at a lower layer. A layer may be able to accept messages of length N, but the implementation of the layer in terms of a protocol at the layer below may accept only messages of length M, with M < N. Thus the original layer must be able to fragment the original message into a number of messages at the lower level. These fragment may then be sent across different routes and the receiver will have to worry about missing fragments, incorrect received order, in recombining them into the correct received message. This problem occurs in particular between the network layer (IP protocol) and the data link layer, where the IP packet needs to be split into a number of fragments. Maximum Segment Size (MSS) is the minimum maximum size of the segments used at different layers in a connection (so if at a layer the maximum segment is 4096 bytes and at another it is 1500, then the MSS is 1500).
The opposite situation may also occur: At the data link layer with Ethernet we have messages of up to 1500 (+26 of overhead) bytes. At the network layer we may have a limit of 128 bytes. [Or the data link unit may be 53 bytes...]
As we need to worry about the maximum sizes in different layers, we need also to worry about the different maximum sizes allowed by the different links that messages will traverse. So one may talk of a path maximum size. If we send packets of at most that size, they will not be fragmented on the way.

End-to-End (E2E) Principle

The original paper. The basic idea is that a communication functionality on a network can be implemented completely and reliably only with the participation of the communicating endpoints. Thus, unless necessary for performance|reliability\security .. reasons, don't put extra intelligence in intermediate nodes (the routers), put the intelligence at the endpoints. That is, the intelligence is kept at the edges.
This principle has had the effect that the internet is fairly simple, basically only a transport and routing functionality. And a consequence of this is that all users of the internet are treated alike, without any particular user, say IBM or Microsoft, receiving special service and thus a competitive advantage. But people now believe that sophisticated processing functions may be appropriate in the internet, not only at the edges, but also in the routers. People have used the term middlebox for these enhanced routers. An example of use of such nodes is in Content Delivery Networks (CDN). Another idea for adding functionality dinamically but safely to the internal nodes, is the concept of isolation kernel. It is a minimal kernel creating an environment on which even thousands of virtual machines can run, each with its own OS that can even crash without affecting anybody else.
Two related papers by David Isenberg are Rise of the Stupid Network and The Dawn of the Stupid Network.
A related concept is Network Neutrality.

Access Networks

The end systems connect to the public internet through Access Networks. These are distinguished into:

Residential Access Network: connects an end system located in a home. It uses mostly the following technologies
- A telephone line (POTS - Plain Old Telephone System - twisted copper pair) and a modem. The data rate, when using standard analog technology and modems, is less than 56Kbps. It is a non shared line. The line from the home to a telephone company's Central Office is called a Local Loop. The place where you reach your Internet Service Provider (ISP) is called a Point Of Presence (POP). In turn the ISP, gets the network facilities, the communication pipes, from Network Service Providers (NSP), companies like Sprint and MCI.
- Using the telephone line, Asymmetric Digital Subscriber Line (ADSL) with a modem. Downstream data rate is more than 1Mbps and the upstream data rate is a few hundreds of Kbps. It is a non shared medium. It includes a separate frequency band to support the traditional phone service.
- Hybrid Fiber Coaxial Cable (HFC) and a cable modem. It is the cable we use for our televisions. It is with different data rates in the two directions: downstream from center to home, less than 8Mbps; upstream from the home, less than 1Mbps. It is a shared medium, so conflicts may arise in the upstream direction.
Institutional Access Network: connects an end system located within an enterprise, say, a university or a business. It uses Local Area Networks (LANs) with the Ethernet technology, 10Mbps Ethernet, 100Mbps Ethernet, even 1Gbps Ethernet.
Mobile Access (Wireless) Networks: connects a movable system to a network. It uses radio frequencies. IEEE 802/11b is a technology for wireless LAns between an access point (a device that is connected someway to the internet) and, through the air, up to 40 cards connected to computers. The distance allowed between access point and cards is < 200 feet, and the total data rate through the access point is 11Mbps. Other technologies you are likely to hear about in this context are Bluetooth, IEEE 802/11a, IEEE 802/11g, and HomeRF.

You will hear of "broadband connection" to mean a connection that is always on, for example when using a cable modem.
You are likely to hear of the Last Mile access problem, the difficulty and cost of bringing the internet to its ultimate users in homes and offices. While fiber optics are being used for long links widely and with ever increasing bandwidths and efficiencies, solutions for the last mile are economically improving at a much lower rate.

Autonomous Systems

The Internet is a collection of Autonomous Systems (AS) which are connected by routers. ASs, in turn, are collections of networks under a single admonistration connected by routers following a common routing strategy. Paraphrasing [RFC1930], an Autonomous System (see also RFC1772) is a set of routers under a single technical administration, using an interior gateway protocol and common metrics to route packets within the AS, and using an exterior gateway protocol like BGP (Border Gateway Protocol, the de facto standard for inter-AS routing [BGP-4]), to route packets to other ASes. BGP sees the ASs organized in a hierarchy, routing packets up and down this hierarchy [communication from source to destination will require moving up to the level of the closest communicating common ancestor(s)]. BGP is concerned with reacheability, not with optimality. Alternatively, an AS is defined as a connected group of one or more IP prefixes run by one or more network operators which has a single well defined routing policy. Where an AS may contain many IP prefixes, an IP prefix should belong to a single AS.

Since these definitions were developed, it has become common for a single AS to use several interior gateway protocols and sometimes several metrics. Even when multiple IGPs and metrics are used, the administration of an AS appears to other ASs to follow a single interior routing plan and presents a consistent picture of what networks are reachable through it.

Autonomous System Numbers (ASNs) are globally unique 16-bit numbers that identify autonomous systems (ASes), and enable an AS to exchange exterior routing information with neighboring ASes.

More information on ASs can be obtained by visiting the American registry for Internet Numbers (ARIN) which assigns the AS numbers, http://whois.arin.net/ or by using the whois program. For example,

    whois -h whois.arin.net temple university

Also interesting is this description of AS interconnectivity. A very good source of information on Autonomous Systems is CAIDA, the Cooperative Association for Internet Data Analysis.

The Internet routers that connect autonomous systems are called AS Border Routers and they exchange routing information [like advertising that to reach particular IP addresses one has to go to go next to particular ASs] using the BGP protocol, version 4. [The routers that actually transmit AS routing information are called speakers. An AS may have many border routers but it has a single speaker.] In this protocol routers collect and exchange using TCP full path information for reaching other autonomous systems and use this information to carry out routing policies (for instance, making the decision to avoid sending traffic through certain ASs) and build routing tables. The information maintained by a router has a Time-To-Live attribute and it becomes obsolete after it expires. AS Border Routers usually are connected by point-to-point links that support high data rates.
Within a particular autonomous system routers communicate using the OSPF protocol (or using the RIP protocol).

There is a hierarchy of Internet providers and their corresponding ASs. Providers at a tier may be peers, i.e. they exchange routing information and forward each other's traffic. Providers at a tier are clients for some provider at the tier above [required to reach the intended destination - it is the default routing destination], and are servers [i.e. the default routing destination] for some providers at the tier below (to be a client of a server in this context may mean having to pay).

A Tier 1 provider (it is a "core router") has one or more specific routes to any node on the Internet, or at least to peer nodes from which any other node can be reached. That is, it can either transport Internet traffic anywhere in the world over its own lines or over those accessible to someone else with which it has a mutual service agreement. A Tier 1 provider is usually treated as a single Autonomous System. Tier 1 providers are also called National Service Providers (NSP). NSPs exchange information through fast public gateways called Network Access Points (NAP) (also called Inter Exchanges or Peering Points) or through private peer-to-peer connections (Bilateral Peering).
Names of some of the tier 1 providers are: - in the USA: Sprint, UUNET (MCI WorldCom), with 30% of the backbone capacity, AT & T, GTE's Internetworking, Global Crossing, Qwest Communications International, PSINet; - internationally: MCI EMEA (NL), Swisscom Enterprise Solutions (CH).

Tier 2 providers are called "regional aggregators". They collect traffic from Tier 3 sites and, if they cannot satisfy them directly, they pass it on to Tier 1 sites. Typically they provide only transport services. A tier 2 provider may also aggreagate IP network addresses.

A Tier 3 provider is the usual "Internet Service Provider" (ISP). ISPs provide transport services and may also provide e-mail and web service.

A Tier 4 provider represents the "backbone LAN" of an organization. It is usually a single autonomous system. Its connection to the outside will go to a tier 3 provider, or, for a sufficiently large organization, directly to a tier 2 provider.

A Tier 5 provider is at the bottom. It is not an autonomous system but one of the LANs that make up such a system.

People distinguish three types of AS:

Stub AS: It is connected to only one other AS. For routing purposes it is treated as part of the parent AS.
Multihomed AS: It is connected to more than one other AS, but does not allow transit traffic. Internally generated traffic can be routed to any of the connected ASs. It is used in large corporate networks that have a number of Internet connections, but do not want to carry traffic for others.
Transit AS: It is connected to more than one other AS and it can be used to carry transit traffic between other AS's.

Tier 1 and Tier 2 providers are usually Transit AS, Tier 3 providers are usually Transit or Multihomed ASs, and the Tier 4 and 5 providers are usually Stub ASs.

Here is a high-level view of the internet, from your workstation, up to the content providers such as yahoo and the New York Times. A good survey of the Internet and of the TCP/IP architecture,

ingargio@joda.cis.temple.edu