π Lecture 76 β Password Authentication Protocol
π Overview: This lecture explains how Internet Service Providers (ISPs) authenticate users connecting via modem or DSL. It contrasts the insecure Password Authentication Protocol (PAP) with the secure Challenge Handshake Authentication Protocol (CHAP), both part of the Point-to-Point Protocol (PPP). Understanding these protocols is essential for network security, as they protect user credentials during remote access.
ποΈ Topics Covered
This lecture covers the Password Authentication Protocol (PAP) and its five-step authentication process using RADIUS or TACACS+. It then details the Challenge Handshake Authentication Protocol (CHAP) and its six-step process that never sends the password over the network. The key difference is that PAP transmits passwords in clear-text, while CHAP uses a message digest computed from a random challenge and the password. The roles of AAA servers, RADIUS, and TACACS+ in handling authentication requests are also explained.
π Lecture Summary
Password Authentication Protocol (PAP)
Password Authentication Protocol (PAP) is a simple authentication protocol used by ISPs as part of PPP (Point-to-Point Protocol) β a data link protocol. PAP sends the username and password in clear-text, meaning anyone with network sniffing tools can read the credentials.
The PAP authentication sequence when Fred dials an ISP with a modem follows these five steps:
- Fred uses PAP to send a username and password to the ISP router.
- The ISP router sends a request to an authentication server using RADIUS protocol messages.
- The authentication server checks the username and password against its database.
- The authentication server confirms that Fred is authentic using RADIUS.
- The router uses PAP to confirm that Fred is allowed to use the Internet.
π Definition β PAP (Password Authentication Protocol): A simple authentication protocol that sends usernames and passwords in clear-text as part of PPP.
π Definition β RADIUS (Remote Authentication Dial-In User Service): A networking protocol that provides centralized authentication, authorization, and accounting (AAA) management for users who connect to a network service.
π Definition β TACACS+ (Terminal Access Controller Access-Control System Plus): A proprietary Cisco protocol developed before RADIUS that can be used in place of RADIUS for authentication services.
π‘ Why this matters: PAP's transmission of passwords in clear text makes it highly vulnerable β anyone with network monitoring tools can capture credentials. This is why CHAP was developed as a secure alternative.
Challenge Handshake Authentication Protocol (CHAP)
Challenge Handshake Authentication Protocol (CHAP) is a more secure alternative to PAP that never sends the password across the network connection. CHAP uses a random number (challenge) and a math function to create a message digest that proves the user knows the password without revealing it.
The CHAP process uses six steps:
- The ISP router generates a random number and sends it in a CHAP message to the user's PC.
- The PC runs a math function using the random number and the password typed by the user as input.
- The PC sends the result of the function β called a message digest β back to the router.
- The router sends the username, the random number, and the message digest to the AAA server. 5a. The AAA server uses the same math function that the client used in Step 2, with the same random number, plus the password associated with that username in the AAA user database. 5b. The result is another message digest. 5c. If the message digest calculated by the AAA server matches the one calculated by the PC, the password that the user typed must be the correct one.
- The AAA server tells the router that the user is authentic; the router tells the PC.
π Definition β CHAP (Challenge Handshake Authentication Protocol): A secure authentication protocol that uses a three-way handshake with a random challenge to verify identity without transmitting the password.
π Definition β Message Digest: The result of running a mathematical function on the combination of a random number and a password β used to prove knowledge of the password without revealing it.
π‘ Why this matters: CHAP is fundamentally more secure than PAP because the password is never passed through a network connection during the entire procedure. An attacker capturing network traffic sees only the random challenge and the digested result β not the password itself.
β Key Takeaways
The critical difference between PAP and CHAP is security: PAP transmits passwords in clear-text making them easily captured, while CHAP uses a random challenge and message digest to verify identity without ever sending the password over the network. Both protocols are part of PPP (Point-to-Point Protocol) used by ISPs. The authentication process involves an ISP router communicating with a central AAA server using RADIUS (or Cisco's proprietary TACACS+). CHAP's six-step process proves password knowledge through mathematical computation rather than transmission, making it the standard for secure remote authentication. For the exam, remember the exact step sequences for both protocols and understand why CHAP is preferred.
π§ Quick Revision Questions
- In PAP, how is the password transmitted to the ISP router?
- What is the purpose of the random number generated by the router in Step 1 of CHAP?
- What is a message digest, and how is it used to verify the user's password in CHAP?
- Name two protocols that can be used by an ISP router to communicate with an AAA server.
- During which step(s) of CHAP does the password travel across the network?
π Lecture 77 β Virtual Private Network (VPN)
π Overview: This lecture explains the concept and benefits of using a Virtual Private Network (VPN). It covers how encryption transforms data into seemingly random bits, allowing secure transmission over the public Internet to create a logically private network.
ποΈ Topics Covered
This lecture covers the fundamental concept of VPNs as encrypted data transmission over the public Internet, how encryption and decryption work with secret keys, the definition of a VPN as a logical private network, and the real-world scenario of an employee using VPN software and a VPN concentrator to connect to a corporate network securely.
π Lecture Summary
Topic 77: Virtual Private Network (VPN)
This topic explains the benefits of using a virtual private network (VPN). Encryption allows a computer to apply a mathematical formula to data, sending the results of the mathematical function over the network. The data just looks like a bunch of random bits. The computer receiving the data can then re-create the original data by decrypting the data. To decrypt the data, you need a secret password called an encryption key.
These days, it is common for users to encrypt data before sending it over the Internet. Instead of calling it encryption, it is called a virtual private network (VPN). The packets go across the Internet, which is still a physical public network. VPNs create a private network, but they do so logically, or virtually.
Letβs assume Barney works for a company. The enterprise network at Barney's company is a private network, with all the components inside privately controlled office space. The Internet is public. For Barney to use the VPN, he must encrypt the packet as he creates it. To do this, Barney needs to have VPN software installed on his computer. He also needs to know the encryption key to use. A VPN device inside the corporate network is called a VPN concentrator, which decrypts packets received from Barney and others. This is shown in Figure 77.1: Encrypting IP Packets for a VPN.
π Definition β VPN (Virtual Private Network): Using encryption to send data securely over a public network (the Internet) to create a logically private connection.
π Definition β Encryption Key: A secret password required to decrypt data that has been encrypted.
π Definition β VPN Concentrator: A device inside a corporate network that decrypts packets received from remote users like Barney.
π Process: VPN Connection β Barney uses VPN software on his computer to encrypt each packet β Encrypted packets travel over the public Internet β VPN concentrator at corporate network decrypts the packets β Original data is recreated for the private enterprise network.
π Example: Barney works for a company with a private enterprise network. He wants to connect securely from a remote location across the public Internet. He has VPN software on his computer and knows the encryption key. When Barney creates a packet, his VPN software encrypts it. The encrypted packet travels over the public Internet to his company's network. The VPN concentrator inside the corporate network receives and decrypts the packet, recreating the original data for the private network.
π‘ Why this matters: VPNs allow remote employees to securely access their company's private network over the public Internet, protecting sensitive corporate data from interception.
β Key Takeaways
The core concept of a VPN is using encryption to create a logically private network over the physically public Internet. Encryption transforms data into seemingly random bits using a mathematical formula, and decryption requires a secret encryption key to restore the original data. A remote user like Barney must have VPN software and know the encryption key to create encrypted packets. The VPN concentrator at the corporate network is the device responsible for decrypting packets received from external users. The key benefit is enabling secure, private communication over a public infrastructure, making it a critical tool for modern remote work.
π§ Quick Revision Questions
- What is the fundamental difference between a physical private network and a virtual private network?
- What two things does a remote user like Barney need to successfully use a VPN?
- What is the name of the device inside the corporate network that decrypts packets from VPN users?
- Why does encrypted data look "like a bunch of random bits" when sent over the network?
- In the VPN process, what travels across the public Internet: the original data or the encrypted data?
π Lecture 78 β Enterprise Network and the Internet
π Overview: This lecture examines what traffic should and should not be allowed between an enterprise network and the Internet. It covers the key considerations when connecting an internal network to the public Internet, including which hosts communicate and which host initiates that communication. Understanding these traffic rules is essential for designing secure enterprise network boundaries.
ποΈ Topics Covered
This lecture discusses the decision-making process for allowing or blocking traffic between an enterprise network and the Internet. It introduces the concept of an Internal IP Network with internal and external web servers, and explains two critical security questions: which two hosts need to exchange packets, and which host begins that communication. The lecture then presents typical types of traffic that are allowed to flow between an enterprise and the Internet.
π Lecture Summary
Enterprise Network and the Internet
When connecting an enterprise network to the Internet, one of the first decisions is what traffic to allow to pass in each direction. Consider an enterprise network called "Internal IP Network" belonging to a company named Fredsco. This network has two types of web servers:
- An internal web server containing content only appropriate for Fredsco employees
- An external web server intended for external users, though internal clients also need to browse it
To secure Fredsco's network, two questions must be answered:
- Between which two hosts do packets need to flow?
- Which host begins that communication?
These questions determine the security rules that govern traffic between the enterprise and the Internet.
π Definition β Enterprise network: A private network belonging to an organization, containing internal resources like servers and client computers.
π‘ Why this matters: Getting these two questions wrong can either leave the network vulnerable to attack or prevent legitimate business communications from functioning.
Typical Types of Traffic Allowed
The lecture presents typical types of traffic that are permitted between an enterprise network and the Internet. These traffic types are determined by the answers to the two security questions above. The figure (Figure 78.2) illustrates common patterns of allowed traffic flows.
π Definition β Traffic flow: The movement of data packets between specific pairs of hosts, with one host initiating the connection.
π Formula: Security decision process: Identify hosts β Determine initiator β Apply allow/block rules
π Example: In the Fredsco scenario, internal clients initiate connections to the external web server, while external users initiate connections to the external web server but should not be able to initiate connections to the internal web server.
β Key Takeaways
The most critical concept from this lecture is that network security between an enterprise and the Internet depends on identifying both the host pairs that need to communicate and which host initiates that communication. For the Fredsco example, internal clients must be able to reach both internal and external web servers, while external users should only reach the external web server. These rules form the basis for configuring firewalls and access control lists. A systematic approach to defining these traffic rules is essential before connecting any enterprise network to the Internet.
π§ Quick Revision Questions
- What are the two questions that must be answered to secure an enterprise network connected to the Internet?
- In the Fredsco example, why must internal clients be able to access both the internal and external web servers?
- Why should external users not be allowed to initiate communication with the internal web server?
- What is the difference between the internal web server and the external web server in terms of intended audience?
- How does identifying which host begins communication affect the security rules?
π Lecture 79 β Firewalls, Demilitarized Zone, IDS
π Overview: This lecture covers the core security mechanisms used to protect networks from Internet threats. It explains how firewalls filter traffic based on rules, the strategic placement of a Demilitarized Zone (DMZ) for public-facing servers, and how Intrusion Detection Systems (IDS) monitor for malicious activity that bypasses the firewall.
ποΈ Topics Covered
This lecture describes the use of firewalls, demilitarized zone (DMZ), and intrusion detection systems (IDS). It explains how a network engineer configures firewall rules to allow or discard packets, the role of the firewall in the network path, and how it identifies new flows using TCP SYN flags. The lecture then introduces the DMZ as a third interface for Internet-accessible servers, with a strict rule against outside-initiated connections to the inside. Finally, it covers network-based and host-based IDS devices that watch packets for signs of cheating or malicious behavior.
π Lecture Summary
Topic 79: Firewalls, Demilitarized Zone, IDS
A network engineer configures the firewall with a set of rules that tells it what's legal and what isn't. The firewall then allows some packets to pass through and discards others to enforce these rules. The firewall must be in the path used for forwarding packets to and from the Internet. Ciscoβs firewall product is called a PIX firewall.
π Definition β Firewall: A security device that enforces a set of rules to allow or discard packets passing between networks. π Definition β PIX firewall: Ciscoβs specific firewall product.
As a firewall watches the traffic entering the network, it knows the nature of the traffic that is allowed to flow through it. It also recognizes when a host is initiating a new flow. A host initiating a new flow can be identified by looking at: The first TCP segment used to create a TCP connection has the SYN flag bit = 1, and the source IP address of the packet. π Example: In Figure 79.2 and 79.3, the firewall sees a TCP segment with SYN=1 from a known source IP. It identifies this as a new flow initiation and checks its rules to allow the TCP connection. π Example: In Figure 79.4, the firewall sees a TCP segment with SYN=1 from an outside host trying to reach an inside web server. The firewall's rule disallows this, so it discards the packet.
π‘ Why this matters: The SYN bit is the key indicator a firewall uses to distinguish between a new connection attempt and an existing flow.
Firewalls are like routers. They forward packets based on destination IP address. They have at least two physical interfaces and can have more than two. The Outside interface connects to the Internet. The Inside interface connects to the internal network. A third interface is connected to a LAN called a demilitarized zone (DMZ) , placed somewhere between the inside and outside interfaces. With a DMZ, Internet-accessible servers can be placed on a different LAN. A stronger firewall rule is applied: No TCP connections can be initiated from outside to the inside. The only flows allowed are for servers in the DMZ.
π Definition β Demilitarized Zone (DMZ) : A separate LAN connected to a third interface on a firewall, used to host public-facing servers like web and email servers, with strict rules preventing outside-initiated connections to the internal network.
Intrusion detection systems (IDSs) watch packets that a firewall allows through. They look for things in the packets to determine if someone is cheating the firewall and doing bad things to servers in the network. Some IDS devices sit in the network, watching packets that pass over a LAN and are called network-based IDSs. Those IDS software that sit on the servers are called host-based IDSs.
π Definition β Intrusion Detection System (IDS) : A security system that monitors packets allowed through the firewall to detect malicious activity. π Definition β Network-based IDS: An IDS device that sits on a LAN and monitors all traffic passing through it. π Definition β Host-based IDS: An IDS software installed on a server to monitor activity on that specific host.
π‘ Why this matters: The IDS provides a second layer of defense, catching threats that the firewall's rules might miss.
β Key Takeaways
The firewall is a rule-based filter that must sit in the network path between the internal network and the Internet. It identifies new connection attempts by looking for the SYN flag set to 1 in the first TCP segment. The DMZ is a separate network segment connected to a third firewall interface, used to host public servers while preventing any TCP connections from the outside to the inside. An Intrusion Detection System (IDS) provides an additional layer of security by inspecting packets that pass the firewall, either on the network (network-based IDS) or on individual servers (host-based IDS). The firewall's core rules typically block all incoming connections from the outside to the inside, only allowing traffic destined for DMZ servers.
π§ Quick Revision Questions
- What specific TCP flag does a firewall look for to identify a host initiating a new flow?
- What is the primary difference between a network-based IDS and a host-based IDS?
- What is the key rule applied to a DMZ regarding TCP connections from the outside?
- What is the name of Ciscoβs firewall product?
- What are the standard names for the two primary physical interfaces on a firewall?
π Lecture 80 β Introduction to Wireshark
π Overview: This lecture introduces Wireshark, a free, open-source network packet analyzer used to capture and display network packets in great detail. It explains the toolβs intended purposes for learning, troubleshooting, security analysis, and debugging, and covers its key features, installation process, and the layout of its main user interface window.
ποΈ Topics Covered
The lecture begins by defining Wireshark as a network packet analyzer and noting its open-source origins with G. Combs. It then lists the intended purposes of the tool for various users (students, administrators, security engineers, developers) and outlines its major features, including cross-platform availability and live packet capture. The installation process is detailed step-by-step, specifying required components like Wireshark, TShark, Plugins, Tools, and the critical WinPcap driver for live capture. Finally, the main window is deconstructed, labeling each area: Title bar, Main menu, Main toolbar, Display filter area, Packet List pane, Packet Details pane, Packet Bytes pane, and Status Bar.
π Lecture Summary
Introduction to Wireshark
Wireshark is a network packet analyzer which captures network packets and displays that packet data as detailed as possible. It is a free open source software program available at www.wireshark.org. G. Combs is the original creator of Ethereal (Wireshark's development name prior to May 2006). As an open source project, its source code is open.
Intended Purposes: When run on a host connected to a wired or wireless network, Wireshark captures and decodes the network frames. People use it to learn network protocol internals. Network administrators use it to troubleshoot network problems. Network security engineers use it to examine security problems. Developers use it to debug protocol implementations.
π Definition β Wireshark: A network packet analyzer that captures network packets and displays detailed packet data. It is free, open source, and runs on UNIX and Windows.
π Definition β Network Packet Analyzer: A tool that captures and decodes network frames for analysis, used for learning, troubleshooting, security examination, and debugging.
Wiresharkβs Features
1- Available for UNIX and Windows. 2- Capture live packet data. 3- Display packets with very detailed protocol information. 4- Filter packets on many criteria. 5- Save packet data captured.
Wiresharkβs Installation
Get Wireshark installer from: www.wireshark.org/download.html and execute it. Figure 80.1 shows Wiresharkβs Installation wizard.
Installation Components include:
- Wireshark - The network protocol analyzer.
- TShark - A command-line network protocol analyzer.
- Plugins & Extensions - Extras for the Wireshark and TShark dissection engines.
- Tools - Additional command line tools to work with capture files.
- Userβs Guide
Figures 80.2 and 80.3 show the component selection screens.
Installing WinPcap
Installing WinPcap is essential. With WinPcap installed you would be able to capture live network traffic. Figures 80.4 and 80.5 show the WinPcap installation process.
π Definition β WinPcap: A driver required for capturing live network traffic on Windows systems. Without it, Wireshark cannot capture live packets.
The Main Window
The main window is shown in Figure 80.6 and consists of several areas:
- Area 1 shows the Title bar β trace file name.
- Area 2 represents Main menu β standard menu.
- Area 3 is the Main toolbar β provides quick access to frequently used items from menu.
- Area 4 depicts Display filter area β reduce amount of traffic you see.
- Area 5 shows Packet List pane β summary of frames.
- Area 6 is the Packet Details pane β dissected frames.
- Area 7: Packet Bytes pane β hex/ASCII details.
- Area 8 represents Status Bar β access to the Expert, annotations, packet counts, and profiles.
β Key Takeaways
Wireshark is a free, open-source packet analyzer for learning network protocols and troubleshooting. It requires WinPcap to capture live traffic. The main window has 8 critical areas: Title bar, Main menu, Main toolbar, Display filter, Packet List pane, Packet Details pane, Packet Bytes pane, and Status Bar. Installation also includes TShark (command-line version), Plugins, Tools, and a Userβs Guide. The tool is available for both UNIX and Windows systems.
π§ Quick Revision Questions
- What is Wireshark and what are its five main features?
- Who created Wireshark (originally called Ethereal)?
- Why is WinPcap necessary during installation?
- What are the eight areas of the main Wireshark window?
- List the four intended purposes of Wireshark mentioned in the lecture.
π Lecture 81 β Five Common Network Problems
π Overview: This lecture identifies and describes five common network problems that can degrade performance or prevent connectivity. Understanding these issues helps network professionals diagnose faults in TCP/IP communications, from connection refusals to slow application responses.
ποΈ Topics Covered
The lecture covers five common network problems: TCP Connection Refused by Server, Connection Blocked by a Host-Based or Network Firewall, Slow Application at Server, Slow Load of Remote Content, and their respective symptoms, causes, and diagnostic indicators in captured traffic.
π Lecture Summary
TCP Connection Refused by Server
This problem occurs when a service is not running on the target server, a firewall blocks the connection, or the destination port is closed. The client initiates a TCP three-way handshake by sending a SYN packet, but instead of completing the handshake, the server responds with a Reset (RST)/ACK packet, refusing the connection. Alternatively, a host-based firewall may respond with an ICMP Destination Unreachable/Port Unreachable message. Possible symptoms include the client's SYN followed by RST/ACK, or the client's SYN followed by an ICMP Port Unreachable response from a host-based firewall.
π Definition β TCP Connection Refusal: A condition where a TCP connection attempt is rejected by the target system because the requested service is unavailable or a firewall blocks the connection. π No formula is provided for this concept. π Example: A client sends a SYN packet to a web server on port 80, but the web server has no HTTP service running. The server responds with an RST/ACK packet, immediately terminating the connection attempt.
Connection Blocked by a Host-Based or Network Firewall
This problem involves a firewall intentionally dropping or rejecting connection attempts to protected resources. Ideally, hosts should not even attempt communication with firewalled resources, but such attempts may occur due to misconfiguration, malware, a malicious user, or other issues. Possible symptoms include no response at all to a SYN packet (the firewall silently drops it), or an RST/ACK response to a SYN packet (the firewall actively rejects the connection).
π Definition β Firewall Block: A security mechanism that prevents unauthorized or unwanted network connections by filtering packets based on rules. π No formula is provided for this concept. π Example: A client attempts to connect to an internal database server from an external network. The network firewall is configured to block all inbound connections to the database port. The client's SYN packet receives no response (silent drop), or an RST/ACK if the firewall is configured to reject the connection.
Slow Application at Server
This problem occurs when the server accepts the connection (no refusal) but responds slowly. The causes include: 1) lack of processing power at the server, 2) a poorly behaving application, or 3) a slow upstream server in a multi-tiered architecture that actually provides the data. The good news is the server provides service; the bad news is the response is delayed. Possible symptoms differ by protocol type: For TCP-based applications, there is a large delay between the server's ACK to a client request and the response data. For UDP-based applications, there is a large delay between a request and the response data.
π Definition β Slow Application: A condition where a server accepts a connection but takes an unusually long time to process and respond to client requests. π No formula is provided for this concept. π Example: A client sends an HTTP GET request to a web server. The server immediately acknowledges (ACK) the request but takes 5 seconds before sending the HTTP response data. This indicates the server-side application is slow to generate the response.
Slow Load of Remote Content
Many networks use a multi-tiered architecture where a client-facing server (Server 1) must obtain data from other servers (Servers 2 through 9) before responding to the client. The client experiences slow loading because the upstream server is slow, not necessarily the front-end server. To diagnose this, network traffic must be captured at the correct point to see which backend server is actually responding slowly. Possible symptoms are the same as Slow Application at Server: for TCP-based apps, a large delay between the server's ACK to a client request and the response data; for UDP-based apps, a large delay between a request and the response data.
π Definition β Multi-Tiered Architecture: A network design where a front-end server depends on multiple backend servers to fulfill client requests, often used in web applications, databases, and enterprise systems. π No formula is provided for this concept. π Example: A client accesses a web page hosted on Server 1. Server 1 must query Servers 2, 3, and 4 to gather dynamic content for the page. If Server 4 is slow to respond, the client sees a slow page load even though Server 1 is functioning normally. Only by capturing traffic between Server 1 and the backend servers can the slow component be identified.
β Key Takeaways
The five common network problems each present distinct symptoms in captured traffic, which network analysts must learn to identify. TCP connection refusals manifest as RST/ACK responses or ICMP Port Unreachable messages immediately after a client's SYN. Firewall blocks may show either no response (silent drop) or active rejection (RST/ACK). Slow application problems at servers and slow remote content loads in multi-tiered environments both present as large delays between request acknowledgment and actual response data. The key distinction between server-side slowness and multi-tiered slowness is that the latter requires capturing traffic at the backend servers to locate the actual bottleneck. Understanding these patterns enables effective troubleshooting using packet analysis tools like Wireshark.
π§ Quick Revision Questions
- What packet(s) would you expect to see in a TCP capture when a server refuses a connection because the requested service is not running?
- What are the two possible symptoms when a host-based or network firewall blocks a connection attempt?
- Name three possible causes for a slow application at the server.
- In a multi-tiered architecture experiencing slow load times, where must you capture traffic to identify the actual slow server?
- How can you distinguish between a TCP-based slow application and a UDP-based slow application from the symptom description?
π Lecture 82 β Next 5 Common Network Problems
π Overview: This lecture covers five additional common network problems that network engineers must diagnose. It explains server application faults, content redirection, buffer issues at both the receiver and sender sides, and problems caused by altered TCP attributes. Understanding these problems is crucial for effective network troubleshooting and performance optimization.
ποΈ Topics Covered
The lecture examines Server Application Fault (where a server is running but not responding to requests), Content Redirection (where applications redirect clients to another host), TCP Receive Buffer Full (where buffer exhaustion stops data transfer), TCP Send Buffer Full (where limited send buffer space degrades performance), and Altered TCP Attributes along a Path (where routers modify TCP connection attributes). Each problem is described with its causes, mechanisms, and observable symptoms.
π Lecture Summary
Server Application Fault
In this scenario, the server is up and running but stops responding to requests. The server successfully responds to a SYN with a SYN/ACK during the TCP handshake, but when the client sends the actual request (e.g., an HTTP GET), no response is received. The client waits for a TCP Retransmission Time Out (RTO), then retransmits the request. If no ACK is received, the client continues to retransmit using an exponential backoff time (doubling the delay each time). Eventually, the client gives up and sends a TCP Reset (RST) to terminate the connection. π‘ Why this matters: This problem can appear as a server being "alive" (pingable) but completely unresponsive to applications, causing user-facing failures.
π Definition β Exponential Backoff: A retransmission strategy where the waiting time increases exponentially (e.g., 1s, 2s, 4s, 8s) between retransmission attempts to prevent network congestion. π Example: A web server accepts the TCP SYN from a client (Step 1), sends SYN/ACK (Step 2), receives the client's ACK (Step 3), but then the client sends an HTTP GET request and receives no response. After 1 second (RTO), the client retransmits the GET. After 2 more seconds, it retransmits again. After 4 seconds, again. The client eventually gives up and sends a TCP Reset (RST).
Content Redirection
This problem is analogous to arriving at a store only to be told the item is out of stock and to go to another store. In networking, if an application supports redirection (like HTTP), and the target server knows where the requested information actually resides, it may send a redirection response to the client. The client then starts a new process to resolve the name of the new host and communicate with it. π‘ Why this matters: This can look like an error or unexpected traffic shift in packet captures.
π Definition β HTTP Redirection: A response from a web server (status codes 300-399) indicating the client should request the resource from a different URL.
π Formula: HTTP Status Codes 300-399 β Client receives "Redirect" and initiates a new name resolution (DNS) and TCP connection to the new host.
π Example: A user requests http://example.com/oldpage, but the server responds with HTTP Status Code 301 (Moved Permanently) and a Location: header pointing to http://newexample.com/newpage. The client then performs a DNS resolution for newexample.com and establishes a new TCP connection to that host.
TCP Receive Buffer Full
This problem occurs when the advertised TCP receive buffer value drops to zero. The receiver uses the Window Size field in TCP headers to tell the sender how much data it can receive. If the buffer is full, the window is set to zero, and data transfer stops completely. Even a low Window Size value (not zero) can significantly slow or stop data flow. The only way to recover is a Window Update packet from the receiver, indicating that buffer space has been freed and the window is now positive. π‘ Why this matters: This causes data flow to stall, resulting in unusually high TCP delays before the transfer resumes.
π Definition β Window Update: A TCP packet sent by the receiver to inform the sender that buffer space is available and data transfer can resume. π Example: A server is downloading a large file from a client. The server's receive buffer becomes full. The server sends an ACK with a window size of 0. The client stops sending data. After the server processes some data and frees up 4 KB of buffer, it sends a Window Update (ACK with window = 4). The client resumes sending data.
TCP Send Buffer Full:
This problem occurs when the send buffer space on the transmitting side is limited. Even if the network has high bandwidth and the receiver has plenty of buffer space, a small send buffer on the sender can bottleneck performance. The sender cannot transmit new data until the send buffer is freed (i.e., until the data it holds is acknowledged by the receiver). π‘ Why this matters: This leads to high TCP delays for no obvious network-related reason, which can be hard to diagnose.
π Example: An application generates data faster than it can be passed to the TCP stack. If the send buffer is only 8 KB, and the sending application writes 16 KB of data, only 8 KB will be sent immediately. The application must wait for TCP to send that 8 KB and receive acknowledgments before it can send the next 8 KB, even if the network and receiver are completely idle and ready.
Altered TCP Attributes along a Path
This problem involves routers and other interconnecting devices that modify the attributes of a TCP connection as they forward the traffic. For example, a router might reduce the Maximum Segment Size (MSS) or alter TCP options in the SYN packets, or it might change the TCP window scale factor. This can cause the sender to calculate a smaller effective window, leading to delays. π‘ Why this matters: This can "break" TCP optimizations and cause performance degradation even on a network with sufficient capacity.
π Definition β TCP MSS (Maximum Segment Size): The maximum amount of data (in bytes) that a TCP segment can contain, excluding the TCP header. It is negotiated during the three-way handshake. π Example: A client and server agree on an MSS of 1460 bytes (typical for Ethernet) during the handshake. However, a router in the path has a lower MTU (e.g., 512 bytes due to a tunnel) and silently truncates the packets or causes fragmentation. This forces the connection to perform at a lower effective segment size, reducing throughput and causing delays before Window Update packets.
β Key Takeaways
- Server Application Fault means the server is alive (accepts TCP connections) but fails to respond to application requests, leading to TCP retransmissions with exponential backoff until a TCP Reset is sent.
- Content Redirection is identified by HTTP status codes 300-399, causing the client to perform a new DNS resolution and connect to a different host.
- TCP Receive Buffer Full stops data flow when the advertised window reaches zero; recovery requires a Window Update from the receiver.
- TCP Send Buffer Full on the transmitting side can halve performance even if the network and receiver are ready, shown by unexplained high TCP delays.
- Altered TCP Attributes along a Path (e.g., changed MSS or window scale) can severely degrade performance, often seen as low calculated window sizes and delays before Window Update packets.
π§ Quick Revision Questions
- What is the sequence of events when a Server Application Fault occurs, from the initial SYN to the final TCP Reset?
- What HTTP status code range indicates content redirection, and what two subsequent network activities does it trigger?
- What specific packet must be sent by a receiver to resume data transfer after its TCP receive buffer is full (window = 0)?
- How can a limited TCP send buffer degrade performance even when the network and receiver have plenty of capacity?
- What TCP attribute is primarily affected when a router in the path has a lower MTU than the endpoints negotiated?
π Lecture 83 β Next 6 Common Network Problems
π Overview: This lecture covers six additional common network problems that can affect network performance and troubleshooting. Understanding these issues is crucial for network engineers to diagnose and resolve connectivity, performance, and data integrity problems effectively.
ποΈ Topics Covered
The lecture describes six common network problems: Mismatched TCP Parameters across a Proxy Device, Routing Loops, Weak Signal (WLAN), Asymmetric Routing, Packet Loss, and High Path Latency. Each problem is defined with its cause and possible symptoms that can be observed during network analysis.
π Lecture Summary
Mismatched TCP Parameters across a Proxy Device
This problem occurs when connection parameters on one side of a proxy device do not match the connection parameters offered on the other side. The proxy device sits between two connections and must translate or negotiate parameters, but mismatches cause queuing delays.
π Definition β Mismatched TCP Parameters: Connection parameters on one side of a proxy device do not match the connection parameters offered in connection on the other side of proxy device. π Possible Symptoms: Delays in data forwarded through proxy (proxy queuing).
Routing Loops
Routing loops occur when a packet is routed back onto a network over and over again, causing it to circulate endlessly until its Time to Live (TTL) expires. This creates duplicate-looking packets that are actually different due to decrementing TTL values.
π Definition β Routing Loop: A situation where a packet is routed back onto a network repeatedly. π Possible Symptoms: Identical packets listed but no TCP Retransmission indications (they are not identical packetsβtheir TTL value decrements). π Example: A packet with TTL=64 enters a loop; each pass decrements TTL to 63, 62, etc. Captured packets look identical except for the decreasing TTL field.
Weak Signal (WLAN)
A weak signal in a Wireless Local Area Network (WLAN) occurs when a wireless signal degrades substantially, potentially causing it to be interpreted improperly when captured by analyzers.
π Definition β Weak Signal (WLAN): A wireless signal that has degraded substantially and may not be interpreted properly when captured. π Possible Symptoms: Low Signal Strength value. π‘ Why this matters: Low signal strength can cause frame errors, retransmissions, and poor throughput, making it a common source of user complaints.
Asymmetric Routing
Asymmetric routing is a situation where traffic flowing from Host A to Host B flows along a different path than the traffic flowing from Host B to Host A. This becomes problematic when devices must see every packet in order to function, such as proxy hosts or Intrusion Detection Systems (IDS).
π Definition β Asymmetric Routing: Traffic flows from Host A to Host B along a different path than traffic flowing from Host B to Host A. π Key Example: A network path must be symmetrical from a client to a proxy host. An IDS box must see every packet to function properly. π Possible Symptoms: ACKed segment that was not captured. π‘ Why this matters: If security devices don't see both directions of traffic, they cannot properly analyze or filter the communication.
Packet Loss
Packet loss typically occurs at an interconnecting device such as a switch, router, NAT device, or network firewall. Small, quick recovery may go unnoticed, but loss of many sequential packets impacts user experience and generates complaints.
π Definition β Packet Loss: Dropping of packets at an interconnecting device such as a switch, router, NAT device, or network firewall. π Possible Symptoms: Fast Retransmission; Duplicate ACKs. π Example: If a router drops 5 sequential packets from a TCP stream, the receiver sends duplicate ACKs triggering fast retransmission at the sender.
High Path Latency
High path latency is caused by a single low-speed (high delay) link along a path or the delay between geographically disbursed peers. This injects a level of path latency that affects overall performance.
π Definition β High Path Latency: A condition where a single low-speed link or geographical distance between peers injects significant delay affecting performance. π Possible Symptoms: Large delays between the outbound SYN and the inbound SYN/ACK of a TCP handshake. π Example: A client in New York connecting to a server in Sydney might experience 200ms+ delay in the SYN-SYN/ACK exchange due to physical distance.
β Key Takeaways
Students must remember the six common network problems: Mismatched TCP Parameters (proxy queuing delays), Routing Loops (identical packets with decrementing TTL), Weak Signal (low signal strength values), Asymmetric Routing (ACKed segment not captured), Packet Loss (fast retransmission and duplicate ACKs), and High Path Latency (large delays in TCP handshake). Each problem has distinct symptoms that appear in network captures, allowing for targeted diagnosis. The routing loop is unique because packets appear identical but differ in TTL, distinguishing them from actual retransmissions. Asymmetric routing primarily affects devices needing bidirectional visibility like IDS. High path latency is specifically identified by delays in the initial TCP handshake phases.
π§ Quick Revision Questions
- What symptom indicates Mismatched TCP Parameters across a Proxy Device?
- How can you distinguish between a routing loop and an actual TCP retransmission in a packet capture?
- What is the primary symptom of a Weak Signal in a WLAN environment?
- Why is Asymmetric Routing problematic for an Intrusion Detection System (IDS) device?
- What specific delay pattern in a TCP handshake indicates High Path Latency?