π Lecture 53 β Router Hardware Architecture
π Overview: This lecture explains the hardware components of a router, which is fundamental to understanding how packets are forwarded in a network. It covers the two key router functions (routing algorithms and forwarding), the architecture of input ports, switching fabrics, and output ports, as well as the critical issue of queuing and packet loss.
ποΈ Topics Covered
The lecture begins by introducing the two key router functions: running routing algorithms (RIP, OSPF) and forwarding datagrams from incoming links to appropriate outgoing links. It then presents a generic router architecture including input ports, switching fabrics, and output ports. The discussion proceeds to explain input port functions, three types of switching fabrics (memory, bus, crossbar), output port functionality, and where queuing occurs at both input and output ports, including the phenomena of Head-of-the-Line (HOL) blocking.
π Lecture Summary
Router Hardware Architecture
There are two key router functions: run routing algorithms (RIP, OSPF) and forwarding datagrams from incoming links to appropriate outgoing links. A generic router architecture includes input ports, a switching fabric, and output ports. The routerβs hardware components work together to perform these functions efficiently.
Input Ports
Input ports perform physical and data link layer functions of an incoming physical link at a router. They determine the output port to which an arriving packet will be forwarded via the switching fabric.
π Definition β Input Port: The router component that performs physical and data link layer functions on an incoming physical link and determines the output port for forwarding via the switching fabric.
Switching fabrics
Switching fabrics transfer packet from input buffer to appropriate output buffer. The rate at which packets can be transferred from inputs to outputs is called the switching rate. The three types are: switching via memory, switching via a bus, and switching via an interconnection network (crossbar).
π Definition β Switching Fabric: The component that transfers packets from input buffers to appropriate output buffers, with a switching rate defining the transfer speed. π Definition β Switching Rate: The rate at which packets can be transferred from inputs to outputs.
Output port
The output port receives packets from the switching fabric and transmits them over the outgoing link. It performs necessary data link and physical layer functions before sending the packets out on the line.
Where does queuing occur?
Packet queues can form at both input and output ports. The location and extent of queuing will depend on the traffic load, speed of switching fabric, and line speed. As these queues grow large, the routerβs buffer space will eventually be exhausted and packet loss will occur when no memory is available to store arriving packets.
Suppose switch speed is N times faster than speed of line. Even then output buffering will still occur when multiple inputs send to the same output. This scenario creates contention at the output port.
The Head-of-the-Line (HOL) blocking at Input Port is a phenomenon where the packet at the front of an input queue is blocked from being forwarded because its destination output port is busy, causing all subsequent packets in that input queue to wait, even if their own destination ports are free.
π‘ Why this matters: Understanding queuing dynamics is critical for designing routers with appropriate buffer sizes and switch speeds to minimize packet loss and latency.
π Definition β Head-of-the-Line (HOL) Blocking: A condition at input ports where the first packet in a queue cannot be forwarded because its destination output port is busy, blocking all subsequent packets in that input queue.
β Key Takeaways
The most critical concepts from this lecture are the two key router functions (routing algorithms and forwarding) and the three main hardware components: input ports, switching fabric, and output ports. Input ports determine the output port for each arriving packet, switching fabrics transfer packets from input to output buffers using one of three methods (memory, bus, or crossbar), and output ports transmit packets over outgoing links. Queuing occurs at both input and output ports, dependent on traffic load, switching fabric speed, and line speed, and can lead to packet loss when buffers overflow. The Head-of-the-Line (HOL) blocking phenomenon at input ports demonstrates how a single blocked packet can prevent other packets in the same queue from being forwarded, creating inefficiency in router performance.
π§ Quick Revision Questions
- What are the two key functions of a router?
- Name the three types of switching fabrics used in routers.
- What determines the location and extent of queuing at router ports?
- Explain what happens when a routerβs buffer space is exhausted.
- Define Head-of-the-Line (HOL) blocking and describe when it occurs.
π Lecture 54 β Routing to Nearby Places
π Overview: This lecture explains how a router learns about networks/subnets that are directly connected to it. When a router's interfaces are configured and operational, it automatically knows which subnets are directly attached and can add those routes to its routing table without needing to communicate with other routers.
ποΈ Topics Covered
The lecture covers how a router discovers directly connected networks once its physical interfaces are up and configured with IP addresses. It explains the process of configuring a new router, how the router uses the IP address and subnet mask to determine the valid range of IP addresses in each subnet, and how it then adds a route for each directly connected subnet to its routing table without needing to forward packets to another router.
π Lecture Summary
Topic 54: Routing to Nearby Places
In this topic, we explain how a router learns routing table for networks/subnets that are directly connected to it.
At a given time, a router knows which of its physical interfaces are up and working. It knows the IP addresses used on each interface. Also, it knows about IP networks or subnets that are connected to its interfaces. The router can add a route to the subnet in its routing table.
Configuring a Router β When you buy a brand new router, a network engineer connects to the router and tells the router which IP addresses to use. For each interface, use an IP Address and Subnet mask to figure out the range of valid IP addresses in each subnet.
After R1's interfaces are up, it knows about subnets: the subnet numbers, the outgoing interface it should use to forward packets to them. There is no need to send packets to another router for them. In our previous example, R1's Ethernet1 interface is connected directly to the same subnet as Hannah. As a result, R1 simply adds a route for each directly connected subnet to its routing table.
β Key Takeaways
A router learns about directly connected networks automatically once its interfaces are configured with IP addresses and subnet masks and are operational. The network engineer's job is to initially configure the IP addresses on each interface. The router uses the IP address and subnet mask to determine the network/subnet number and the range of valid IP addresses. For directly connected subnets, the router adds a route with the outgoing interface but no next-hop router, because packets can be sent directly. This is the simplest way a router populates its routing table, and it forms the foundation for learning routes to remote networks through other routing protocols.
π§ Quick Revision Questions
- What information does a router need to know about each of its interfaces to discover directly connected subnets?
- How does a router determine the subnet number and valid IP address range for a directly connected network?
- When a router has a directly connected route, does it need a next-hop router address in its routing table? Why or why not?
- What is the role of a network engineer when configuring a new router?
- In the example with R1 and Hannah, why does R1 not need to send packets to another router to reach Hannah's subnet?
π Lecture 55 β Dynamically Learning Routing Tables
π Overview: This lecture explains how routers learn routing table entries dynamically through routing protocols, rather than relying solely on static configuration. It covers the fundamental process by which routers exchange route information, enabling them to build comprehensive knowledge of network topology automatically.
ποΈ Topics Covered
The lecture begins by contrasting static routes with dynamic learning, then introduces routing protocols as the mechanism for automatic route discovery. It explains the step-by-step process of how routers initially know only directly connected routes, then generate and exchange routing updates containing network/subnet information. Finally, it details how receiving routers process these updates by recording the incoming interface and the sending router's IP address as the next-hop.
π Lecture Summary
Topic 55: Dynamically Learning Routing Tables
This topic explains how routing table entries are learnt dynamically. The most typical way a router learns all the rest of the routes in an internetwork, beyond just its directly connected routes, is by using a routing protocol. Routing protocols define messages by which routers can exchange route information with other routers. If all the routers participate, all routers should have routes for all subnets or networks.
Letβs assume an internetwork that consists of two routers and three subnets. Several steps that occur over time are:
- Each router knows only its directly connected routes.
- A router generates a routing protocol message (called a routing update) that contains information about IP networks and subnets and sends it to another router.
- The receiving router: a) Puts that interface into the route in which it received the routing update. b) Also puts the IP address of the router that sent the routing update as the next-hop router.
The lecture illustrates this with a diagram (Figure 55.1) showing R2 advertising its routes, with R1 learning a route to subnet 150.1.3.0. In the process:
π Definition β Routing Protocol: A set of messages and rules that routers use to exchange route information with each other, allowing dynamic learning of network topology.
π Formula: Route learned = Destination subnet + outgoing interface (where update received) + next-hop router (IP address of sending router)
π Example: Consider an internetwork with two routers (R1 and R2) and three subnets. Initially, R1 knows subnet 150.1.2.0 (directly connected on Ethernet1) and R2 knows subnet 150.1.3.0 (directly connected on Ethernet2). R2 sends a routing update containing "I have a route to subnet 150.1.3.0". R1 receives this update on its Ethernet1 interface (interface where update arrived). R1 records: destination = 150.1.3.0, outgoing interface = Ethernet1, next-hop = R2's IP address (150.1.2.2). Now R1 can forward packets to 150.1.3.3 via R2.
π‘ Why this matters: Dynamic routing eliminates the need for manual configuration of every route, making networks scalable and resilient to topology changes.
β Key Takeaways
Routing protocols are the essential mechanism for dynamic route learning, allowing routers to automatically exchange routing information beyond their directly connected networks. The learning process follows a consistent three-step pattern: initial knowledge of only directly connected routes, generation of routing update messages containing known network/subnet information, and processing of received updates where the receiving router records the incoming interface and the sender's IP address as the next-hop. This dynamic approach solves the problem of static routing, particularly the difficulty of adapting to network topology changes and the inability to utilize multiple physical paths simultaneously. The critical distinction is that while directly connected routes are automatic when an interface is configured and working, all other routes must be learned either statically (configured by an engineer) or dynamically (via routing protocols). Successful routing protocol implementation requires all routers to participate, ensuring every router has complete knowledge of all subnets and networks in the internetwork.
π§ Quick Revision Questions
- What three pieces of information does a receiving router record when it processes a routing update?
- Why do static routes make it difficult to utilize multiple physical paths to the same destination?
- What is the initial routing knowledge of each router before any routing protocol exchange occurs?
- What is the purpose of a routing update message in a routing protocol?
- How does the dynamic learning process ensure that routers eventually have routes for all subnets in an internetwork?
π Lecture 56 β How to Pick the Best Route
π Overview: This lecture explains how routing protocols select the best path to a destination when multiple routes exist. It introduces the concept of a metric β a numerical value that determines route quality β and describes how routers react when a preferred route fails. Understanding best-route selection is fundamental to building reliable and efficient networks.
ποΈ Topics Covered
The lecture begins by describing a three-router internetwork where Router R1 can reach subnet 150.1.3.0 through either R2 or R3. It then explains how routing protocols use a number called a metric to evaluate each route's quality. The scenario concludes with a failure event: when the Ethernet switch between R1 and R2 loses power, the routing protocol removes the failed route from the routing table, leaving only the backup route through R3 active.
π Lecture Summary
How the Best Route Gets Selected
Routing protocols not only help routers learn routes, but they also help routers learn the best routes to a destination when there is more than one way to get there. Assume an internetwork consisting of three routers. R1 can send packets to subnet 150.1.3.0 through R2 or through R3.
![Figure 56.1: How the best route gets selected]
π Definition β Metric: A number defined by a routing protocol that is associated with each route in a routing update. That number represents how good or how bad that route is.
When a router receives multiple routing updates, it might learn of multiple ways to reach a subnet. By looking at the metrics, the router can pick the best route.
![Figure 56.2: The router can pick the best route]
Handling Route Failure
Imagine that someone turns off the power on the Ethernet switch between R1 and R2. That route would then fail and be unavailable. The routing protocol on R1 would remove the route from the routing table. At the same time, R1's routing protocol would continue to learn about the alternative route through R3, so packets destined for subnet 150.1.3.0 would be forwarded through that remaining route.
π‘ Why this matters: Route failure handling ensures network resilience. When the primary path fails, the routing protocol automatically removes the dead route and relies on the surviving backup route, maintaining connectivity without manual intervention.
β Key Takeaways
Routing protocols serve a dual purpose: they learn routes and also identify the best route when multiple paths exist. The metric is the decisive factor β it is a numerical value assigned to each route that indicates its quality or cost. When multiple routes to the same destination are learned, the router selects the one with the best metric. If a route fails (for example, because a switch loses power), the routing protocol removes that route from the routing table. As long as an alternative route remains available through another path, the router continues to forward traffic using that surviving route.
π§ Quick Revision Questions
- What is the purpose of a metric in a routing protocol?
- In the example with subnet 150.1.3.0, through which two routers could R1 reach the destination?
- What does a router do when it receives multiple routing updates for the same subnet?
- What happens to a route when the Ethernet switch between two routers loses power?
- After the Ethernet switch between R1 and R2 fails, how does R1 forward packets to subnet 150.1.3.0?
π Lecture 57 β Interior & Exterior Routing Protocols
π Overview: This lecture explains why routing on the internet cannot be done with a single, flat routing protocol and introduces the concept of hierarchical routing through Autonomous Systems (ASes). It distinguishes between interior routing protocols (used within an AS) and exterior routing protocols (used between ASes), which is fundamental to understanding how the global internet scales and respects administrative boundaries.
ποΈ Topics Covered
The lecture covers the limitations of flat routing (scale and administrative autonomy), the definition and role of Autonomous Systems (ASes), the function of gateway routers, and the distinction between intra-AS (interior) and inter-AS (exterior) routing protocols. It uses a diagram to illustrate how gateway routers within one AS forward packets to destinations outside that AS.
π Lecture Summary
Hierarchical Routing
Considering the network to be flat, taking all routers to be identical and running one single routing protocol is not realistic. There are at least two reasons:
- Scale: With hundreds of millions of destinations, it is impossible to store all destinations in routing tables. Similarly, routing table exchanges would swamp links.
- Administrative Autonomy: The internet is a network of networks. Each network admin may want to control routing in its own network.
π‘ Why this matters: Without hierarchy, the internet would collapse under the weight of its own routing information, and no single organization would accept another controlling its internal routing decisions.
π Definition β Autonomous System (AS): Organize routers into regions. An AS is under the same administrative control. For example, operated by the same ISP or belonging to the same company network.
All routers within the same AS run the same routing protocol (intra-AS or interior routing protocol) and have information about each other. Routers in different ASes can run different intra-AS routing protocols.
π Definition β Gateway routers: To connect ASes to each other, one or more of the routers in an AS will have to be responsible for forwarding packets to destinations outside the AS.
π Example: In the diagram, if a router in AS1 receives an IP packet destined outside of AS1, then it uses an inter-AS (exterior) routing algorithm to learn which destinations are reachable through which gateway router, and then forwards the packet accordingly.
β Key Takeaways
A flat routing model is unrealistic due to scale and administrative autonomy, so the internet is organized into Autonomous Systems (ASes). Each AS is under a single administrative control and runs its own intra-AS (interior) routing protocol among its routers. Gateway routers connect different ASes and use inter-AS (exterior) routing protocols to learn reachability to destinations in other ASes. This hierarchical structure allows the internet to scale globally while respecting the autonomy of individual networks.
π§ Quick Revision Questions
- What are the two main reasons why a single, flat routing protocol across the entire internet is not realistic?
- Define an Autonomous System (AS) and give an example of what entity might operate one.
- What is the role of a gateway router in the context of connecting different ASes?
- Do routers in the same AS always run the same interior routing protocol? Do routers in different ASes always run the same interior routing protocol?
- When a router in AS1 receives a packet destined for a host outside AS1, what type of routing algorithm does it use to decide where to forward the packet?
π Lecture 58 β Introduction to Domain Name System
π Overview: This lecture explains why a Domain Name System (DNS) is required for the Internet. It covers the need for multiple identifiers, the function of DNS as a distributed database, and the core services DNS provides. Understanding DNS is essential for grasping how user-friendly hostnames are translated into router-friendly IP addresses, making Internet communication seamless.
ποΈ Topics Covered
The lecture introduces the concept of multiple identifiers for Internet hosts, contrasting human-friendly hostnames with machine-friendly IP addresses. It explains the need for a Domain Name System as a distributed database, analogous to a phone book. The hierarchical namespace structure is described, along with the roles of Top-Level Domains (TLDs), Second-Level Domains (SLDs), and subdomains. Finally, the core services provided by DNSβhostname-to-IP-address translation, host aliasing, mail server aliasing, and load distributionβare detailed.
π Lecture Summary
Why is DNS Required?
People have many identifiers (e.g., name, passport#, NTN#, NIC#). One identifier cannot be used everywhere. Similarly, Internet hosts have many identifiers. A hostname (e.g., www.yahoo.com) is human-friendly but difficult for routers to process variable length alphanumeric characters. An IP address (32 bit) is router-friendly. If you need to call someone and you know the person's name but not the phone number, you can look up the information in a phone book. TCP/IP hosts can have the equivalent of a phone book in a file. The local host file contains a list of TCP/IP host computer names and their corresponding IP addresses.
π Definition β Hostname: A human-friendly alphanumeric identifier for a network host, such as www.example.com. π Definition β IP Address: A 32-bit numeric identifier that routers use to deliver packets across networks. π Example: If Hannah types http://www.example.com in a browser, the system needs to find the corresponding IP address to establish a connection.
DNS is a Distributed Database
When the Internet was small, a single hosts.txt file (managed by the Network Information Center at SRI) was used. An administrator would email changes, and the updated file was downloaded weekly. This method had obvious drawbacks: the file and traffic would be huge for the growing Internet, host names could not be unique, and the process was inefficient.
DNS is not a single database but is distributed across many servers. It is implemented in a hierarchy of many name servers. DNS is an application-layer protocol β the client-server paradigm runs over UDP on port 53. DNS is also a core Internet function, meaning if DNS does not work, many other services (web, email, file transfer) also stop working.
π Definition β Name Server: A server that stores DNS database records and responds to DNS queries. π Definition β Application-Layer Protocol: A network protocol that operates at the application layer (Layer 7) of the OSI model, used for communication between programs. π Formula: DNS Operation β User β Query β DNS Server β Response (IP Address) β User connects to destination.
DNS Namespace
The DNS namespace is organized hierarchically into a tree structure (like a directory tree). The root of the tree is unnamed. The root is a logical concept β there are 13 logical root name servers (each replicated for redundancy). Under the root are Top-Level Domains (TLDs).
There are three types of TLDs:
- Generic Top-Level Domains (gTLDs): .com, .org, .net, .edu, .gov
- Country Code Top-Level Domains (ccTLDs): .pk, .uk, .au
- Infrastructure Top-Level Domain: .arpa (used for reverse DNS lookups)
Under TLDs are Second-Level Domains (SLDs) (e.g., foo.com is a SLD under .com). Each SLD can have subdomains (e.g., eng.foo.com, sales.foo.com). Each subdomain can be further divided. A domain is a subtree of the DNS namespace.
π Definition β Namespace: A hierarchical tree structure that organizes all domain names on the Internet.
π Definition β Top-Level Domain (TLD): The highest level in the DNS hierarchy, such as .com, .org, .edu.
π Example: In the hostname eng.foo.com, the sequence is: . (root) β .com (TLD) β .foo (SLD) β .eng (subdomain).
DNS Services
DNS provides several essential services beyond simple name-to-address translation:
- Hostname to IP Address Translation: This is the primary service. DNS resolves a hostname (e.g., www.yahoo.com) into an IP address (e.g., 68.180.206.184).
- Host Aliasing: A host can have multiple names. The canonical hostname (actual name) can have one or more alias hostnames. DNS resolves an alias to the canonical name.
- Mail Server Aliasing: Email addresses use simple names like
user@company.cominstead of the actual mail server name. DNS resolves the alias to the canonical mail server name. - Load Distribution: Busy sites (e.g.,
cnn.com) are replicated across multiple servers with different IP addresses. DNS translates the same hostname into different IP addresses for different requests, distributing the load.
π Definition β Canonical Hostname: The actual, primary hostname of a server. π Definition β Alias Hostname: An alternative name for a server that points to the canonical hostname. π‘ Why this matters: Load distribution via DNS is a simple and effective way to balance traffic across multiple web servers without requiring complex hardware load balancers at the application layer.
β Key Takeaways
The lecture establishes that DNS is an indispensable core Internet service that translates human-friendly hostnames into machine-friendly IP addresses. Unlike the early days of a single hosts.txt file, DNS is a distributed, hierarchical database that scales globally. Its namespace is structured as a tree from root to TLDs to SLDs and subdomains. Beyond basic translation, DNS provides critical services including host aliasing, mail server aliasing, and load distribution. A deep understanding of DNS is fundamental because its failure impacts virtually all Internet applications β web, email, and file transfers all depend on DNS resolution.
π§ Quick Revision Questions
- Why can't we simply use a single, centralized hosts.txt file for name resolution on the modern Internet?
- What are the three main categories of Top-Level Domains (TLDs)? Give an example of each.
- Explain how DNS performs load distribution for a busy website like www.cnn.com.
- What is the difference between a canonical hostname and an alias hostname?
- On which transport layer protocol and port does DNS typically operate?
π Lecture 59 β Domain Name System (DNS) Servers
π Overview: This lecture explains the need for multiple domain name system (DNS) servers organized hierarchically, addressing the scalability problems of a centralized DNS. It describes how different types of DNS servers interact to resolve a domain name into an IP address, and introduces the roles of root, TLD, authoritative, and local DNS servers.
ποΈ Topics Covered
The lecture begins by introducing DNS as an application-layer protocol for name resolution, then outlines the major problems with a centralized DNS approach: single point of failure, traffic volume, distant database delays, and maintenance burdens. It presents the solution of a distributed, hierarchical DNS server architecture, explaining the interaction flow between root servers, top-level domain (TLD) servers, authoritative servers, and local DNS servers. A concrete example of resolving www.amazon.com is used to illustrate this process.
π Lecture Summary
Domain Name System (DNS) Servers
DNS is an application-layer protocol that allows hosts to query a database and to resolve names (name, address). It is employed by other application-layer protocols such as HTTP, SMTP, and FTP.
Problems with centralized DNS
A centralized DNS approach presents four critical problems:
- Single point of failure: If the DNS server crashes, so does the entire Internet.
- Traffic volume: A single server would have to handle all DNS queries.
- Distant centralized database: A single server cannot be βclose toβ all the querying clients β causing significant delays.
- Maintenance: A centralized database would have to be updated frequently for every new host.
π Definition β Centralized DNS: A single server holding all name-to-IP mappings for the entire Internet.
In order to deal with the issue of scale, the DNS uses a large number of servers, organized in a hierarchical fashion and distributed around the world.
Interaction of Servers: an approximation
Letβs assume a client wants the IP address for www.amazon.com.
- The client queries a root server to find the com DNS server.
- The client queries the com DNS server (TLD) to get the amazon.com DNS server.
- The client queries the amazon.com DNS server (Authoritative) to get the IP address for www.amazon.com.
π Definition β Root Name Servers: In the Internet there are 13 root servers.
π Definition β Top-level Domain (TLD) Servers: Responsible for com, org, net, edu, gov, and all top-level country domains, e.g.: uk, fr, ca, jp.
π Definition β Authoritative Servers: Organizations with publicly accessible hosts (e.g. web servers) must provide publicly accessible DNS records that map the names of those hosts to IP addresses.
π Definition β Local DNS Servers: Do not strictly belong to the hierarchy. Each ISP has one server which is also called a βdefault name serverβ. When a host makes a DNS query, the query is sent to its local DNS server.
π‘ Why this matters: The hierarchical and distributed design of DNS servers prevents the Internet from having a single point of failure, distributes query traffic globally, and allows updates to be managed locally by the organization that owns a domain. Without this structure, the Internet as we know it could not scale.
β Key Takeaways
A centralized DNS is impractical due to single-point-of-failure, massive traffic volume, geographic delays, and frequent maintenance overhead. The actual DNS uses a distributed, hierarchical architecture with root servers, TLD servers, authoritative servers, and local DNS servers. There are exactly 13 root servers in the Internet. The resolution process for www.amazon.com involves three sequential queries: first to a root server to locate .com, then to the .com TLD server to locate amazon.com, and finally to amazon.com's authoritative server to get the IP address. Local DNS servers, one per ISP, act as the first point of contact for client queries and cache results to improve performance.
π§ Quick Revision Questions
- What are the four problems with a centralized DNS approach?
- How many root name servers exist in the Internet?
- What is the role of a Top-level Domain (TLD) server?
- Who provides authoritative DNS servers for a domain like amazon.com?
- What is the function of a local DNS server, and does it strictly belong to the DNS hierarchy?
π Lecture 60 β Working of Domain Name System
π Overview: This lecture explains the working of the Domain Name System (DNS) both inside and outside a company. It covers how DNS resolves names to IP addresses within an enterprise network and how it handles external website requests, including the concept of DNS caching.
ποΈ Topics Covered
This lecture covers the internal DNS resolution process within a company's enterprise network, illustrated with the example of Hannah's PC accessing a local website. It then explains the external DNS resolution process when accessing a website hosted in a different company's network.
π Lecture Summary
Working of Domain Name System
This topic explains how the Domain Name System (DNS) functions to figure out names and IP addresses for the entire Internet, as well as inside a single site at a single company.
Inside the Company
Assume that Hannahβs PC is inside the example.com corporation's enterprise network and she wants to view http://www.example.com.
For DNS to work well inside a company, someone must be responsible for supporting it. That work includes updating and changing the list of names and IP addresses.
Outside the Company
Assume that Hannah wants to access the http://www.fredsco.com website, which is located in the Fredsco enterprise network.
π Definition β DNS Caching: once (any) name server learns a mapping, it caches that mapping for future use.
π Formula: No formula is presented in this lecture.
π Example: The lecture uses two contrasting scenarios to illustrate DNS working. In the first scenario, Hannah (inside example.com) accesses a website within her own company's network. In the second scenario, she accesses a website located in a different company's network (fredsco.com), highlighting that the resolution process involves servers outside her own enterprise.
β Key Takeaways
A student must understand that DNS operates differently for internal and external name resolution within an enterprise network. DNS requires a responsible administrator to maintain the list of names and IP addresses inside a company. The critical concept of caching is introduced, where name servers store learned mappings to improve efficiency. The lecture provides two concrete network diagrams showing how a single user's request travels depending on the target domain's location.
π§ Quick Revision Questions
- What is the role of an administrator in making DNS work well inside a company?
- According to the lecture, what happens when a name server learns a mapping?
- What is the difference between Hannah accessing www.example.com versus www.fredsco.com?
- What term describes the process where a name server stores a learned mapping for future use?
- In the context of this lecture, what does DNS define?
π Lecture 61 β DNS Resource Records
π Overview: This lecture explains the structure and use of DNS Resource Records (RRs), which are the fundamental data units stored in DNS servers that map domain names to IP addresses. It describes the different types of resource records and demonstrates the practical process of registering a new domain name and how DNS queries resolve to allow web browsing.
ποΈ Topics Covered
This lecture covers the structure of DNS Resource Records including the four key fields (Name, Value, Type, TTL), the five main type definitions (A, NS, CNAME, MX, AAAA), and a step-by-step walkthrough of how domain names are registered with ICANN-accredited registrars. It then follows the complete DNS resolution process when a user attempts to access a newly registered website.
π Lecture Summary
DNS Resource Records
The DNS distributed database is stored on servers in the form of resource records (RRs). Each resource record contains four fields: (Name, Value, Type, TTL). The Type field defines how the Name and Value fields should be interpreted.
The Type field is defined as:
- Type = A β Provides a standard hostname to IPv4 address mapping: (hostname, IP address)
- Type = NS β Specifies an authoritative DNS server for a domain: (domain, hostname of authoritative name server)
- Type = CNAME β Provides canonical name alias: (alias hostname, canonical hostname)
- Type = MX β Specifies mail exchange server: (alias domain name, canonical name of mail server)
- Type = AAAA β Provides a hostname to IPv6 address mapping: (hostname, IPv6 address)
Inserting resource records into DNS
Letβs assume that Alice just created a new startup company βNetwork Utopiaβ. She wants to register domain name networkuptopia.com.
π Definition β Registrar: An organization accredited by the Internet Corporation for Assigned Names and Numbers (ICANN) that verifies the uniqueness of domain names, enters them into the database, and collects a small fee for its services.
Step 1: Alice contacts a registrar to register her domain name. Step 2: She provides the registrar with the names and IP addresses of her primary and secondary authoritative name servers. Step 3: The registrar inserts two RRs into the TLD com server:
- Type NS record: (networkutopia.com, dns1.networkutopia.com, NS)
- Type A record: (dns1.networkutopia.com, 212.212.212.1, A) Step 4: Alice must ensure that:
- A Type A record for her web server www.networkuptopia.com is entered into her authoritative DNS servers
- A Type MX record for her mail server mail.networkutopia.com is entered into her authoritative DNS servers
π‘ Why this matters: This registration process is what makes your website visible to the entire internet β without these RRs properly configured, no one can find your server.
The DNS Resolution Process
Now suppose Bob views www.networkutopia.com:
- Bobβs host sends a DNS query to his local DNS server
- The local DNS server contacts a TLD com server
- The TLD com server contains the NS and A RRs listed above and sends a reply back to Bobβs local DNS server
- Bobβs local DNS server then sends a query to 212.212.212.1, asking for the Type A record
- The authoritative name server responds with the Type A record, providing the IP address of the web server, say 212.212.71.4
- Bobβs host now has the IP address and the browser can initiate a TCP connection
π Example: When Bob types www.networkutopia.com, his local DNS server queries the TLD com server, which responds with the NS record pointing to dns1.networkutopia.com (212.212.212.1). The local server then queries that authoritative server for the Type A record of www.networkutopia.com and receives IP 212.212.71.4, enabling Bob's browser to open a TCP connection.
β Key Takeaways
The DNS system relies entirely on Resource Records which store mappings between names and values with specific type definitions. There are five critical RR types: A (IPv4 address), NS (authoritative name server), CNAME (alias), MX (mail server), and AAAA (IPv6 address). Registering a domain requires working with an ICANN-accredited registrar who inserts NS and A records into the TLD server, while the domain owner must configure A and MX records on their own authoritative servers. The resolution process involves a chain: local DNS server β TLD server β authoritative name server, with each step querying for specific RR types. Understanding this flow is essential for diagnosing DNS failures in real-world networks.
π§ Quick Revision Questions
- What are the four fields contained in every DNS Resource Record?
- What is the difference between a Type A record and a Type NS record?
- Which organization accredits domain registrars, and what is the registrar's role?
- List all five Type definitions discussed in this lecture and briefly explain each.
- Trace the complete DNS resolution process when a user tries to access www.networkutopia.com for the first time.
π Lecture 62 β Introduction to Wide Area Network
π Overview: This lecture introduces the concept of a wide area network (WAN) and explains how it differs from local area networking. It highlights why simple Ethernet cabling cannot be used over long distances due to legal and physical constraints, setting the foundation for understanding dedicated WAN technologies.
ποΈ Topics Covered
The lecture defines what a wide area network (WAN) is in terms of device distance and legal cabling boundaries. It then presents a simple scenario of connecting two routers with an Ethernet crossover cable, explaining how signal transmission works between them. Finally, it identifies two critical problemsβlegal and physicalβthat prevent using Ethernet over long distances like between Atlanta and Cincinnati.
π Lecture Summary
What is a Wide Area Network?
A wide-area network (WAN) defines a type of network, or part of a network, in which the devices are relatively far apart. A WAN is a network, or part of a network, for which the cabling must pass outside the property of one company. The distance might only be a few miles, or it might be thousands of miles!
How Two Routers Could Connect (In Theory)
You can run an Ethernet cross-over cable between two devices and the two devices can communicate with Ethernet. That works well, but can we use it for longer distances? If two routers in two cities need to forward packets to each other, they need some sort of physical medium over which to send the packets.
When R1 sends out an electrical signal over the cable, R2 needs to receive that same signal on the wires that it expects to receive data. Likewise, R1 needs to receive what R2 transmits. Because the physical interfaces on the routers use the same pins to transmit, the cable connects the twisted pair used for transmitting by R1 over to R2's receive pins, and vice versa. In short, all that the two routers need is a cable between them, with transmit and receive pairs of wires, so that they can send and receive anytime they want.
π Definition β wide-area network (WAN): A network or part of a network in which the devices are relatively far apart and the cabling must pass outside the property of one company.
Problems with Using Ethernet over Long Distances
In the previous example, there are two problems that stop you from using Ethernet:
- 1: You are not legally allowed to run a cable between Atlanta and Cincinnati.
π‘ Why this matters: Even though two routers could theoretically communicate using a simple Ethernet crossover cable, the physical and legal barriers of running cables across public or third-party property make Ethernet impractical for WAN connections. This is why dedicated WAN technologies and service providers are necessary.
β Key Takeaways
A WAN is defined by both distance and the fact that cabling must leave a company's property, unlike a LAN. While two routers can theoretically communicate using an Ethernet crossover cable with proper transmit/receive wiring, legal restrictions prevent running cables across cities or states. The core problem is not technical feasibility but legal permission to use the physical medium over public or third-party land. This lecture sets the stage for understanding why service providers and dedicated WAN technologies exist.
π§ Quick Revision Questions
- What is the definition of a wide area network (WAN) according to this lecture?
- How does an Ethernet crossover cable allow two routers to send and receive signals to each other?
- What is one specific problem that prevents using Ethernet between Atlanta and Cincinnati?
- Does the lecture state that Ethernet is technically incapable of working over long distances?
- Why must WAN cabling pass outside the property of one company?
π Lecture 63 β Different Aspects of WAN Link
π Overview: This lecture explains how telephone companies (telcos) enable long-distance communication between two routers located miles apart, such as Atlanta and Cincinnati. It covers the physical and logical services required to create a WAN link, including the role of serial interfaces, 4-wire circuits, and central office switches.
ποΈ Topics Covered
The lecture covers the three main services a telco must provide to connect two distant routers: a cable-like service with crossed wire pairs, proper physical encoding via serial interfaces, and the infrastructure of central offices (COs) with WAN switches that ensure the path between locations. It emphasizes the telco's existing cable networks and legal right-of-way for installation.
π Lecture Summary
To Connect Two Routers (e.g., Atlanta and Cincinnati, 450 miles)
A telephone company (telco) must offer a service that acts like a cable with four wires (two pairs). When a router sends data on one pair, that pair is crossed to the other pair before reaching the other end. This way, when one router sends on one pair, the other router receives on the other pair. Routers use a serial interface because the service requires physically encoding data a little differently than on Ethernet. The result is that two routers can send and receive data to and from each other. Essentially, a telco can lease a 4-wire cable or 4-wire circuit between two points.
π Definition β 4-wire circuit: A physical cable with four wires (two pairs) leased by a telco to connect two distant routers, enabling bidirectional data transmission.
Telco Infrastructure
Telcos have already run cables between almost every town and city. They have offices called central offices (COs) almost everywhere. In addition, telcos have the right-of-way, which is the legal right to dig up roads and put cabling in the ground.
π Definition β Central Office (CO): A telco office that houses switches and connects local cables to long-distance networks.
π Definition β Right-of-way: The legal permission granted to telcos to dig up roads and install underground cabling.
Switches and Path Establishment
Switches in the CO are called telco switches, phone switches, or WAN switches. The telco must ensure a path between Atlanta and Cincinnati by providing:
- A cable from the office building in Atlanta where one router sits to the telco CO in Atlanta.
- A cable from the office building in Cincinnati where the other router sits to the telco CO in Cincinnati.
β Key Takeaways
The most critical points from this lecture are that telcos provide a 4-wire circuit that acts like a crossed cable, enabling two distant routers to communicate using serial interfaces. Telcos have pre-existing cable networks connecting almost all towns and cities, with central offices everywhere and legal right-of-way to install new cabling. WAN switches in the COs route traffic between locations. Serial interfaces are required because the physical encoding differs from Ethernet. Understanding the telco's role in providing physical infrastructure and path establishment is fundamental to grasping WAN link operation.
π§ Quick Revision Questions
- What three main services must a telco offer to connect two distant routers?
- How does the crossing of wire pairs enable bidirectional communication between two routers?
- Why do routers use a serial interface instead of an Ethernet interface for a WAN link?
- What is the legal right-of-way, and why is it important for telcos?
- Name the type of switch found in a central office that helps establish the path between two distant locations.