Computer networks

Lecturer:Gabinet Artem Viktorovych

Lecture 1. Course overview.

date: 2016-02-18 08:30:00 +0200

Course intro

Network components:

  • Devices
  • End devices – devices we use for smth (smartphones, servers, laptops, etc)
  • Intermediate devices – devices which allows us to pass data (commutators, routers, etc)
  • Medium
  • Messages – pass message from one device to another, big messages are splitted to small pieces and being sent separately
  • Rules describe how devices pass messages to each other.

Lecture 2. Network design. Network models.

date: 2016-02-25 08:30:00 +0200

Network design

There are 4 characteristics of well-designed network. Here they are:

  • Redundancy
  • Scalability – Network have to be easy to extend witout modifying existing parts of the network
  • QaS (Quality of Service) – Crucial data must have priority over other data
  • Security

In the future lectures network topology illustrations will be used.

Legend:

End devices: [] Generic, [L] Laptop, [S] Switch

Intermediate devices: [==] Switch, (+) Router, \(+)/ Wifi router

When messages are too big, they are split into smaller parts. This is called multiplexing. Packets are enumerated. This allows to restore a whole message from parts

[]--------[==]--(+)
         / |
        /  |
[]-----/   |
           |
           |
 LAN       []

Network types:

  • LAN – Local Area Network
  • WAN – Wide Area Network
  • Internet

Network models

Standard developers:

  • IEEE Institute of electric and electronics engineering
  • IETF Internet engineering task force

Protocols are described in documents called RFC

TCP/IP Model

№ Name
4 Application
3 Transport
2 Internet
1 Network access

OSI Model (developed by ISO)

№ Name
7 Application
6 Presentation
5 Session
4 Transport layer
3 Network
2 Data link
1 Physical
  • Application layer is the layer which is used by user applications
  • Presentation layer– data encoding, compressing, etc
  • Session layer – ??
  • Transport layer – transferring data between applications
  • Network layer – transfering data between devices
  • Data link – interface between application and hardware layers
  • Physical layer – hardware level

Each protocol has its own PDU (Protocol data unit).

Level(s) PDU
7, 6, 5 Data
4 Segment
3 Packet
2 Frame
1 Bytes

Lecture 3. Physical level.

date: 2016-03-10 08:30:00 +0200

OSI levels from 2 (Data link) to 7 (application) are software levels. Developers use them in their software.

The first level (physical) is the trully hardware level.

Physical level generates either electrical/radio/other signals.

Physical level takes incoming frame and encodes it in form of ones an zeros. Then it generates corresponding signals to send this data.

Physical level responsibilities include:

  • Data encoding/decoding
  • Signaling

Physical level encodings:

NRZ

NRZ
|--+   +-----+
|  |   |     |
|  |   |     |      Uses clocking. One bit sent in one bittime (time unit)
+--+---+-----+-----
  1  0   1  1  0

NRZ was used in really slow connections. There is a problem, when several same values come in a sequence (like 4 ones), because it is pretty difficult for receiver to “understand” this data.

Manchester coding (the first one)

Manchester coding (the first one)
| +--+   ++ +--+
| |  |   || |  |
| |  |   || |  |
+-+--+---++-+--+------
  1  0   1  1  0

Change from no signal to existing signal is 1 and from existing signal to no signal is 0. This was much effective than NRZ, but still worked only for slow connections

4B/5B

Every 4bits are encoded in special combination of 5 bits. That allows to avoid issues with several same values in sequence.

Disadvantage: data overhead

8B/10B

The same as 4B/5B, but instead of 4bit and 5bit sequences, 8bit and 10bit are used instead.

Bandwidth

Bandwidth shows how much data can be sent per some time interval. It is measured in bps (bits per second)

1000bps  = 1 kbps
1000kbps = 1 Mbps
1000Mbps = 1 Gbps
1000Gbps = 1 Tbps

Bandwidth of Ethernet is 10Mbps

Bandwidth depends on both transmitter and receiver. If transmitter and receiver have different bandwidth, the lowest of them is used.

  1Gbps  10Mbps  1Gbps
[]-----[]------[]-----[]

Bandwidth in above picture is 10Mbps

Throughput and goodput

Throughput is the data transmission speed at specific conditions.

max:100Mbps  max:100Mbps
[]------{cloud}-----[]
  60Mbps       100Mbps

Throughput in above picture is 60Mbps

Goodput is the payload transmission speed at specific conditions. That means, techncal and service data (frame, package, segment, etc) headers are not counted in goodput.

Connections

  • Twisted pair
  • Fiber connection
  • Wireless
  • Coaxial

Twisted-pair

Twisted-pair has two standards:

  • UTP unshielded twisted pair.
  • SPT shielded twisted pair.

Max length of Twisted-pair is 100m

Twisted pair consists of 8 wires.

  • Orange
  • Green
  • Blue
  • Brown

There are 2 standards:

  • T568A
  • T568B
  T568A T568B
1 Green/ Orange/
2 Green Orange
3 Orange/ Green/
4 Blue Blue
5 Blue/ Blue/
6 Orange Green
7 Brown/ Brown/
8 Brown Brown
Straightthrough twisted-pair: A-A or B-B
Crossover twisted-pair: A-B or B-A

Straighthrough:

  • Switch + PC
  • Switch + Router

Crossover:

  • PC + PC
  • Router + Router
  • Switch + Switch

MDI-x allows us to not bother about which twisted-pair (Straightthrough or crossover) we use.

Only 4 wires are used to transfer data at 100Mbps. Other 4 are used to power devices

Fiber

2 types:

  • Multimod (d = 50-60mkn), up to 1.5km
  • Singlemod (d = 9-10mkn), uses laser, up to 80km

Lecture 4. Fiber. Wireless. Application level.

date: 2016-03-17 08:30:00 +0200

Fiber

Fiber cable usually uses two fiber wires: one wire to transmit data in one direction

There are one-wire fiber cables. These cables use light waves with different wave length

Two types of connector:

  • LC FC
  • SC

Ports

  • GBIC - old format
  • SFP (1 Gbps)
  • SFP+ (10 Gbps)
  • QSFP (40Gbps)
  • QSFP+ (80Gbps)

Connection speed can be increased using link aggregation: two cables on the software level recognized as one link, and balancer sends part of packets on the one cable, and part – on another cable.

Wireless

802.11 standard.

standard name frequency speed
standard 802.11a 2.4Ghz up to 11 Mbps
standard 802.11b 5Ghz up to 52 Mbps
standard 802.11g 2.4Ghz up to 52 Mbps
standard 802.11n 2.4Ghz, 5Ghz up to 300 Mbps
standard 802.11ac 2.4Ghz, 5Ghz > 1 Gbps

2.4 Ghz is used by wide variety of devices, hence if there is another device which works on this frequency nearby, the major loss of quality will occur due to signal interference.

There are three channels which do not overlap at 2.4Ghz frequency. These are 1st, 6th and 11th channels.

Application level

Main protocols:

  • HTTP
  • DNS
  • DHCP
  • FTP
  • telnet
  • SSH
  • IMAP
  • POP3
  • SMTP
Client/Server architecture
Peer-to-peer (P2P)

FTP

Data transfered in a plain form. No encryption.

DHCP

Dynamic host configuration protocol.

  1. Client broadcasts request to receive IP address (DHCP-discover)
  2. DHCP server finds accessible ip addresses and returns DHCP-offer.
  3. Client receives DHCP-offer and broadcasts its intent to use offer. (DHCP-Request)
  4. Corresponding DHCP server registers client and returns DHCP-Acknowledge response to client
  5. Client saves received IP address

Lecture 5. Application level protocols. Telnet. SSH. HTTP. DNS.

date: 2016-03-24 08:30:00 +0200

Common protocols

DHCP, FTP are text protocols

Telnet

Was developed for remote access to other device’s console.
Uses client-server architecture

All logins, passwords data are passed in unencrypted way.

SSH

Secure SHell

Was developed for remote access to other device’s console.
Uses client-server architecture

All logins, passwords data are passed encrypted.

HTTP

HyperText Transfer Protocol

Client-server architecture

Client  Webserver
  [L]-------[S]
HTTP protocol up to version 1.1 (inclusive) is text protocol.
HTTP 2.0 is binary

HTTP request according to protocol consists of 3 parts

  • Request string (mandatory)
    Request string consists of
    Request type Resource identifier Protocol version
    GET/POST/PUT/PATCH/DELETE/… in general it is URI i.e HTTP/1.1

    Example request string: GET http://google.com.ua/ HTTP/1.1

  • Request headers (optional)

  • message body (optional)

HTTP protocol is stateless. Hence workaround is required to store state. Such workaround is called Cookies.
Cookies are passed as request headers. Usually they are encoded in base64

DNS

Domain Name Service

This protocol is used to resolve request url into actual IP address.

Domain name system is a distributed structure. On the top of this structure root domain-name server is placed. This one knows how to resolve top level of domain name. On the next level first-level domain servers are placed. They “know” where to find second-level domain servers and so on and so forth.

              [root (.)]
     +------+----+---+------+-----+
   [com]   [ua]    [edu]  [org]  [io] ...
  +--+--+
  |     |
[vk]   [github]...

Fully qualified domain name ends with dot: fiot.kpi.ua.

Domain name is resolved from end to beginning. DNS server delegates request to the corresponding lower-level DNS server.

If we want to open mlp.wikia.com:

  1. DNS resolution request to DNS server to associate domain name with IP address.
    1. request [.] for mlp.wikia.com. response [com] DNS server
    2. request [com] for mlp.wikia.com. response [wikia.com] DNS server
    3. request [wikia.com] for mlp.wikia.com. response [mlp.wikia.com] IP address
Types of DNS records:

  • A. Domain name is associated with IP: foto IN A 77.77.77.77

  • PTR associates IP address with domain name. in-addrarpa.
    (77.47.128.130)
    in-addrarpa
 |  .... \ ...
77
 |
47
 |
128
 |
130

  • TXT.

  • SOA. (Self-off authority). Describes DNS server responsibilities (authority)

  • NS. name server. kpi.ua IN NS ...

  • MX (Mail eXchange).

DNS lookup can be either interactive (DNS Server delegates to another DNS server) or recursive (Server recursively searches for response without delegating)

Примітка

nslookup tool =)

Mail Protocols:

     SMTP           IMAP/POP3
[L]------------[S]-------------[PC]
sender      mail server       receiver

IMAP

This one is used to receive emails. This protocol is newer than POP3.
IMAP loads only mail headers. And concrette message will be loaded as requested.

POP3

This one is used to receive emails. It loads whole email messages at once.

SMTP

This protocol is used to send email.

Lecture 6. Transport layer.

date: 2016-03-31 08:30:00 +0200

Transport level

PDU: Segment

  • data exchange between applications. It does not matter, where the application is, it only matters which application. So transport level does not transmits data between hosts, it only determines which data belongs to which application and passes data from/to corresponding application.
  • splits data into small parts (chunks)
  • determines, which application received data belongs to
  • message tracking
  • flow controll.

Message passing on transport level

  1. Ordering received messages
  2. Reliability. Wait for transmission confirmation.
  3. Flow controll. Message will be transfered in batches (several parts transmitted at a time)
  4. Segmentation.

Protocols

Dependeing on protocol specific header is added to message.
In both TCP and UDP application identifier is passed in header. 2b is used to store this application id. Hence there are 2^16 unique identifiers (ports).
Ports from 0 to 1023 are called well-known
Ports from 1024 to 49151 are called registered ports. Use these ports when you develop your app
Ports from 49152 to 65535 are Dynamic ports
Protocol Port Transfer protocol
FTP 20,21 TCP
SSH 22 TCP
TELNET 23 TCP
SMTP 25 TCP
DNS 53 UDP (when client requests data), TCP (to sync DNS servers)
DHCP 67,68 UDP
HTTP 80 TCP
POP3 110 TCP
IMAP 143 TCP

Within header both source and destination ports are passed.

 SRC PORT   DST PORT       DATA
+---------+----------+-------------+
|  50000  |    80    |             |
+---------+----------+-------------+

TCP. Transmission controll protocol

  • TCP protocol guarantees data transmission.
  • Header size: 20b
|16b   SRC PORT|16b   DST PORT|
|6b:flags|                    |
|32b: ISN                     |
  • Before data transmission source and destination establish TCP connection (3-wave handshake):
|                      |
|------SYN,seqa------->|
|                      |
|<---SYN,ACK(seqb)-----| If there is nothing on this port RST returned instead
|                      |
|----ACK(seqb + 1)---->| Now we have TCP connection established
|                      |
|                      |
|                      |
|---DATA(1602)-------->| Data size passed within seq param. 1602 is size here
|                      |
|<---ACK(1603)---------| Data transfer ackknowledge.
|                      |
|                      |
|-----FIN------------->| A sent all data and wants to finish connection
|<----FIN--------------| B wants to finish connection too
|-----ACK------------->| Confirm connection end
A                      B

flags: SYN, ACK, RST, FIN, PUSH, URG

  • TCP tries to send more data with every sequential request. I.E. within first transfer it sends 1kb. Then 2, 4,8,16 and so on and so forth. (flow control)

    use netstat

UDP. User datagram protocol.

  • Does NOT guarantee data transmission
  • Header size: 8b

Lecture 7. Network layer. IPv4

date: 2016-04-07 08:30:00 +0200

Network layer

In addition to data encapsulation and decapsulation which occur on every layer, network layer is also responsible for addressing devices and routing.

Network layer protocols include

  • IPv4
  • IPv6

Network layer is fully unreliable. You cannot be sure if the data will be sent/received.

Network layer protocols are unaware of data transmission method (wifi, ethernet, smth else) used at physical layer.

Nets and subnets

    [ ]-  -[ ]
        \/
 [ ]----------[ ]
        /\
    [ ]-  -[ ]

This how it used to be earlier
But now there are so many devices that having one net for all of them is not enough.
Here subnets are coming for rescue.

There are different ways of distributing subnets (or splitting net into subnets)

  • geographical. This is the dummiest way. I.E. one floor in 10-floor building has its own subnet.
  • purpose. Some people (finance dpt) need high reliable conection for work and other (managers) does not. We can create subnets for every group of peopple with specia configuration
  • security. Sometimes for security reasons we need to restrict access for different groups of devices. Subnets for the rescue.
   [ ]-              -[ ]
       \            /
 [ ]---[=]---(+)---[=]---[ ]
       /            \
   [ ]-              -[ ]

Now we have two subnets =)

Gateway is used to communicate between subnets.

Addressing

Sample IPv4 address: 193.47.196.75. IPv4 address’ length is 32bits (4 octets). It consists of 4 bytes splitted with dots.

IP address consists of two parts:

  • Network part
  • Host part
193.47.196.75  /24
----------|----|----
Network   |Host|Prefix
part      |part|
Prefix determines how many bits are used for network part
To determine IP address of network we have to fill host part of IP address with zeros.
To get broadcast address we have to fill host part of IP address with ones. (bits)
Address type IP address
Host IP 192.47.196.75/24
Network IP 192.47.196.0/24
Broadcast IP 192.47.196.255/24

Number of available hosts in network is:

$$ N = 2^n - 2 $$

Here

  • $N$ is the number of available hosts in network
  • $n$ is the length of the host part in bits $$n = 32 - prefix$$

IPv4 address classification:

| Class | IP address range           |Prefix| Note                           |
|-------|----------------------------|------|--------------------------------|
|   A   |  1.0.0.0 -- 127.255.255.255| /8   | These networks were given to   |
|   B   |128.0.0.0 -- 191.255.255.255| /16  | big organizations as there are |
|   C   |192.0.0.0 -- 223.255.255.255| /24  | a lot of available hosts       |
|-------|----------------------------|------|--------------------------------|
|   D   |224.0.0.0 -- 239.255.255.255|      | Multicast addresses            |
|   E   |240.0.0.0 -- 255.255.255.255|      | Reserved addresses             |

There are some reserved addresses:

  • 127.0.0.0/8 – localhost
  • 169.254.0.0/16 – M$ addresses used in networks w/o DHCP
  • 192.0.52.0/24 – WTF is this?
  • 0.0.0.0
  • 255.255.255.255 – network broadcast address

Addresses can be either public or private. Private addresses are used in private networks and are unaccessible from Internet.

Private addresses lay in the following ranges:

  • 10.0.0.0/8
  • 172.16.0.0/12 – 172.31.255.255/12
  • 192.168.0.0/16

NAT

NAT stands for Network address translation. This topic will be covered after test will be passed.
This is used to give Internet access for devices which obtain private IPv4 address

Message types:

There are three main types of messages:

  1. Unicast. Only one host is the recipient of mesage.
  2. Broadcast. Message is sent to every available host in the broadcast domain. One broadcast domain is equal to one subnet. Broadcast messages do not pass through layer-3 devices.
  3. Multicast. Multicast messages are sent to specified group of devices.
  4. Anycast. Messages are routed to the topologically nearest node in a group of potential receivers, though it may be sent to several nodes, all identified by the same destination address.
  5. Geocast. Message is sent to a group of destinations in a network identified by their geographical locations. It is a specialized form of multicast addressing used by some routing protocols for mobile ad hoc networks.

Lecture 8. Subnets. Test preparations.

date: 2016-04-14 08:30:00 +0200

Subnets

IANA organization is responsible for distributing IPv4 addresses. It used to be a single organization, but nowadays it has transformed into group of organizations:

  • ARIN (North America)
  • APNIC (Asia)
  • AFRINIC (Africa)
  • RIPENCC (Europe)
  • ACNIC (Australia))

Splitting network into subnets:

Given: 193.175.16.0 / 24

In binary format: 11000001.10101111.000100000.00000000 / 24

Assume, we want to make $$n = 2$$ subnets:

Increase network part by $$\lceil log _{2}(n) \rceil = 1$$ : 11000001.10101111.000100000.0 | 0000000 / 25

Now we can have 2 subnets. Here they are:

  • 11000001.10101111.000100000.0 | 0000000 / 25
  • 11000001.10101111.000100000.1 | 0000000 / 25

Rewrite addresses in decimal:

  • 193.175.16.0 / 25
  • 193.175.16.128 / 25

Assume, we want to split 193.175.16.128 / 25 into $$n = 4$$ subnets:

Increase network part by $$ \lceil log _{2}(n) \rceil = 2 $$ : 11000001.10101111.000100000.100|00000 / 27

Now we can have 4 subnets. Here they are:

  • 11000001.10101111.000100000.100 | 00000 / 27
  • 11000001.10101111.000100000.101 | 00000 / 27
  • 11000001.10101111.000100000.110 | 00000 / 27
  • 11000001.10101111.000100000.111 | 00000 / 27

Rewrite addresses in decimal:

  • 193.175.16.128 / 27
  • 193.175.16.160 / 27
  • 193.175.16.192 / 27
  • 193.175.16.224 / 27

Planning network

Given:

  • 193.175.16.0 / 24

  • Five departments:

    DPT PCS prefix network
    A 27    
    B 100    
    C 15    
    D 4    
    E 30    

  1. Determine subnet prefixes according to number of PCs:

    DPT PCS prefix network
    A 27 /27  
    B 100 /25  
    C 15 /27  
    D 4 /29  
    E 30 /29  

    TODO: Describe this process

  2. Split network into subnets:

    TODO: Describe this process

  3. Assign subnets to department

    TODO: Describe this process

Test preparations

Task 1

Task Description

Given: 177.250.13.246 / 28

Find: Network, Netmask, Broadcast address, 1st and last host IP in network:

Solution
  1. Write down IP in binary form: 101110001.11111010.00001011.11110110
  2. Determine host and network parts: 101110001.11111010.00001011.1111 | 0110
  3. Net: fill host part with zeros: 101110001.11111010.00001011.1111 | 0000
  4. Netmask: fill network part with ones and host part with zeros: 111111111.11111111.11111111.1111 | 0110
  5. Broadcast: fill host part with ones: 101110001.11111010.00001011.1111 | 1111
  6. First available IP: 101110001.11111010.00001011.1111 | 0001
  7. Last available IP: 101110001.11111010.00001011.1111 | 1110

Finally transform binary results to decimal:

  1. Net: 177.250.13.240
  2. Netmask: 255.255.255.240
  3. Broadcast: 177.250.13.255
  4. First available IP: 177.250.13.241
  5. Last available IP: 177.250.13.254
  6. Total IP addresses available: $$ 2 ^ {32 - 28} - 2 = 14 $$

Task 2. Split network into subnets.

NOTE: I’m not sure, I got it right. Please correct me if I made a mistake, otherwise remove this note if you can.
Task Description

Given:

  • 10.1.14.0 / 24

  • departments:

    dpt PCs prefix subnet
    1 13    
    2 120    
    3 15    
    4 61    
    5 2    

To do: Divide to subnets and assign subnets.

Solution

  1. Determine prefixes. I.E. For 120:

    • Keep in mind that you have to add 2 to number of PCs, as one address will be used as network address and other one will be used for broadcast.
    • Nearest bigger power of two is 128.
    • Thus we need $$ log _{2}(128) = 7 $$ bits for the host part
    • And finally calculate the network prefix: $$ prefix = 32 - 7 = 25 $$
    dpt PCs prefix subnet
    1 13 /28  
    2 120 /25  
    3 15 /28  
    4 61 /26  
    5 2 /30  

  2. Divide net into subnets:

    • 10.1.14.0 / 25
    • 10.1.14.128 / 25
      • 10.1.14.128 / 26
      • 10.1.14.192 / 26
        • 10.1.14.192 / 28
        • 10.1.14.208 / 28
        • 10.1.14.224 / 28
        • 10.1.14.240 / 28
          • 10.1.14.240 / 30
          • 10.1.14.244 / 30
          • 10.1.14.248 / 30
          • 10.1.14.252 / 30

  3. Assign subnets:

    dpt PCs prefix subnet
    1 13 /28 10.1.14.192 / 28
    2 120 /25 10.1.14.0 / 25
    3 15 /28 10.1.14.208 / 28
    4 61 /26 10.1.14.128 / 26
    5 2 /31 10.1.14.240 / 30

Lecture 9. Routing

date: 2016-04-28 08:30:00 +0200

One of the Network Layer responsibilities is routing.
Among the data, which is sent, destination IP and source IP are passed over network. This is done to make possible determine sender and reciever. Transport protocol code is also passed in package headers. To make sure that header is OK, CRC checksum is passed in that header (recursion, yup)

ICMP protocol

Internet control message protocol. This protocol is used for network diagnostics. It works over the Network layer. One of the most well-known features of thisprotocol is ability to check node availability.

TTL – time to live
Every 3rd level device (like router) decrease TTL by one. If TTL value is 0, then device drops the packet. Initial TTL value is set by sender. That depends on OS.

RTT – Round trip time

[]---[==]-(+)---(+)---(+)---[]

Routing

WTF is router? (+)

Router is a box with holes, buttons, wires and light indicators. It is some sort of computer, without monitor. There are CPU, RAM, NVRM, ROM, physical interfaces, etc…

Router is a 3rd level device. That means, it works on 3rd (Network) layer.

Despite everything else router allows us to unite different networks with each other, find best path to some host, and so on and so forth.

Routing table

Every router stores routing table. Routing table stores records of networks, next-hop or interfaces, where the network is.

There are 3 types of records in routing table:

  1. Directly connected. It appears as soon as router is enabled or device is plugged into network/router
  2. Static routes. System administrators/users fill these records.
  3. Dynamic routes. These records are populated by routers themselves, depending on their configurations.

Rules of routing

  1. If you have a route on one router in netwrok, it does not mean other ruoters have that route.
  2. If you create a route, you have to create a backward one
  3. Next-hop must be in directly-connected network

Static routes

Попередження

I’ve messed upo this topology. PLS FIX IT!

                192.168.2.0/24
                      ___
                       |
                       |
R1. 192.168.4.0/30     |192.168.5.0/30        192.168.3.0/24
   (+)i2------------i1(+)i2-----------------------(+) R3.
 i1+.254              R2.                          |
  /                                              [==]----[]
\\
 192.168.1.0/24

R1

192.168.1.0/24 | i1
192.168.4.0/30 | i2
192.168.3.0/24 | 192.168.4.2

192.168.3.0/24 next-hop 192.168.4.1

R2

192.168.4.0/30 | i1
.2.0/24 | i3
.5.0/30 |
.3.0/24 | 192.168.5.2

Lecture 10. Dynamic routing

date: 2016-05-12 08:30:00 +0200

                R2
            ----(+)-| B
      R1   /     |
A |---(+)--      |
           \     |
            ----(+)-| C
                R3

There are some running processes on routers. They exchange some data, which allow to build routing tables for networks.

Chief weaponry of dynamic routing protocols include such diverse things as

  • Messages
  • Path finding algorithms
  • Corresponding data structures

Comparisson of static and dynamic routing

Criteria Static Dynamic
Scaleability Bad: network must be reconfigured manually to scale Good
Knowledges Little knowledge A lot of knowledge required
CPU, resourses Minimal Uses CPU to configure routing
Reconfiguration Complex Simple
Security Good Bad
Fault tolerance Extremely bad Good (depends on used protocol)

Routing configuration protocols can be divided into two groups:

  • IGP Interior gateway protocol
  • EGP Exterior gateway protocol

Autonomic system is a group of networks running under a single administrative control. This could be our company or a branch of company. Just like Subnetting AS is also used to break a large network in smaller networks.

Classification

  • EGP
  • BGP
  • IGP
  • distance-vector
  • link-state
    • IS-IS (Intermediate system to intermediate system)
    • OSPF (Open shortest path first)

Distance-vector protocol know nothing but the next hop and some metric, called distance.

Link-state protocols try to build a full network topology graph and then find paths on that graph.

RIP =(

  A   R1   B  R2   C  R3   D
|-----(+)-----(+)-----(+)-----|
R1 R2 R3
A - 0 B - 0 C - 0
B - 0 C - 0 D - 0
———– ———– ————
C: R2 - 1 D: R3 - 1 B: R2 - 1
  A: R1 - 1  

Every 30 seconds routers broadcast their routing table to their “neighbours”.

Metrics:

Metric Protocol
  • Hop count
 RIP
  • Bandwidth
  • Delay
  • Load
  • Reliability
 EIGRP
  • Cost
 OSPF

If there are several paths with the same metrics, RIP (as well as any other protocol) will balance traffic among these pathts.

Maximum diameter of network, to which RIP can be applied is 15

Split horizont. If router R2 received path to D from R3, it will not send recieved path (D) to R3.

Lecture 12. Cool stuff

date: 2016-05-12 08:30:00 +0200

DHCP

     [ ] DHCP
      |
      |
    [===]----------(+)----{ INTERNET }
    / | \
   /  |  \
[ ]  [ ]  [ ]
As we know, DHCP requests do not pass through third-layer devices.
But what if we have a topology like this?
        DHCP  [ ]---[===]---[ ] WEB
                      /
                     /
    [===]---------1(+)----{ INTERNET }
    / | \            2
   /  |  \            \
[ ]  [ ]  [ ]        [===]
                     / | \
                    /  |  \
                 [ ]  [ ]  [ ]

To handle this we could use own DHCP server for each subnetwork, which is extremely inconvenient and expensive (A lot of subnets == A lot of expensive server hardware). How can we handle this?

  • Setup DHCP Relay on router ports 1 and 2!!!
  • Setup default gateway for router

Now the router will repack BROADCAST DHCP requests into UNICAST requests and pass them to DHCP server.

  • Setup IP pools for DHCP server:
{
    "pool1": {
        "for network": "192.168.1.0/24",
        "def gateway": "192.168.1.254",
        ...
    },
    "pool2": {
        "for network": "192.168.2.0/24",
        "def gateway": "192.168.2.254",
        ...
    },
    ...
}

And voila! If you did everything correctly, your DHCP server will serve to several subnets as well as it does to one!

If your DHCP server goes down, your whole network goes down. To handle this you may want to configure DHCP Failover

NAT

IPv4 has too little IP addresses to assign to devices. There are even much less public addresses. However there are a lot of devices, which need to have an IP address. So how can we workaround this?

     [ ] DHCP
      |
      |              77.47.10.1
    [===]------.254(+)------------{ INTERNET }
    / | \                              |
  .1  .2 .3                          [   ] WEB
[ ]  [ ]  [ ]                          vk.com (1.1.1.1)

192.168.1.0/24

NAT stands for Network Address Translation. There are 3 Types of NAT:

  • Static
  • Dynamic
  • Masquarade

Static NAT

If router recieves request to Internet from 192.168.1.1 then it translates address to 77.47.11.1 We’ll have the following header:

Source Dest Data
77.47.11.1 1.1.1.1 Data

When router recieves request from outer (like 1.1.1.1) to 77.47.11.1 it translates destination IP to corresponding Internal address (192.168.1.1)

All translations are saved in NAT table.

Static NAT is not really good, because it requires to have public address for each device.

Dynamic NAT

Instead of setting correspondence between addresses it sets correspondence between pools of addresses and public address. If several requests from pool are received, they are translated in sequential order.

Masquarade NAT (PAT)

PAT stands for Port Address Translation.

All requests from (i.e. .254) port of router are translated into one public IP address.
I.E. we are sending request from .1 PC:
  SRC DST
IP 192.168.1.1 1.1.1.1
Port 33101 80

Then we’ll have the following record in NAT table:

Local IP Local Port Global IP Global Port DST IP DST Port
192.168.1.1 33101 77.47.10.1 20001 1.1.1.1 80

If router will recieve another request (even to the same resource) from 192.168.1.2 during request it’ll add another record to NAT table and perform a corresponding substitute

Local IP Local Port Global IP Global Port DST IP DST Port
192.168.1.1 33101 77.47.10.1 20001 1.1.1.1 80
192.168.1.2 54104 77.47.10.1 20002 1.1.1.1 80

When response is recieved, corresponding record will be deleted from table.

What happens inside NAT stays inside NAT. No internal activity is visible from outside.

But what if we need to access some internal resource from outside? Port Forwarding is used for that. You configure NAT so it translates all incoming requests to specific port (i.e. 8080) to specific device (i.e. 192.168.1.3)

Firewall

Firewall is used to filter requests according to some rules. I.E. you may drop all requests to 8080 port both incoming and outcoming.

Interestings

You can use Wireshark to discover and explore your network traffic. This application allows you to analyse incoming and outcoming packages. Have fun =)