Common - Question 4 (Networks) Flashcards
Layers of ISO/OSI and TCP/IP
Application
Presentation
Session
Transport
Network
Data Link
Physical
Application
Transport
Internet
Network Access
Main IPv4 protocols
ARP (Address Resolution Protocol): Maps IP addresses to MAC addresses as
a part of the interface between the network and link layer.
ICMP (Internet Control Message Protocol): Network layer protocol generating
control messages, often in response to errors in IP operations. Each message has
a type (e.g., Destination unreachable), a numerical code and a checksum.
DHCP (Dynamic Host Configuration Protocol): A client/server protocol
that automatically provides an IP address and related configuration information
such as the subnet mask and default gateway to a host.
Also RARP, IGMP
Network requirements
- efficiency – efficient/maximal use of available throughput
- fairness – the same approach to all the data flows (having the same priority)
decentralised management - fast convergence when adapting to a new state
- multiplexing/demultiplexing
- reliability
- data flow control – a protection in order to avoid network’s (network devices’) and hosts’ congestion
Define “network protocol”
Network Protocol is a set of rules that defines the format and the order of messages exchanged among two or more communicating entities, as well as the actions performed during sending/receiving that messages
L1 main functionality (OSI/OSI physical layer)
Bit-to-signal transformation, bit rate control, bit synchronization, multiplexing, circuit switching
L2 Data Link layer functionality (ISO/OSI model)
Framing, addressing, error control, flow control, medium access control
Medium access protocols examples
Random access: Aloha, CSMA/CD (Collision detection), CSMA/CA (Collision avoidance)
Services of L3 - Network layer
Internetworking - ‘gluing’ physical networks
Packetizing
Fragmenting - MTUs
Addressing - IP addresses
Address resolution - ARP, RARP protocols
Routing
Control messaging - ICMP protocol
L4 Transport layer services
Packetizing, connection control, addressing (using ports), connection reliability (flow and error control), congestion control, quality assurance.
Name advanced IPv6 features
Larger address space
Extension headers
Path MTU discovery
Neighbor discovery protocol
Mobility support for devices
Security (IPSec)
What is routing
Routing - the process of finding a path in the network between two communicating nodes
Name families of routing protocols with examples
Distance Vector:
- RIP (Routing Information Protocol)
- E/IGRP (Enhanced Interior Gateway Routing Protocols)
Link State:
- OSPF (Open Shortest Path First)
- IS-IS (Intermediate System to Intermediate System )
Path Vector:
- BGP (Border Gateway Protocol)
Describe Distance Vector protocols
the neighboring routers periodically (or when the topology changes) exchange complete copies of their routing tables
Uses Bellman-Ford algorithm
Describe Link State protocols
the routers periodically exchange information about states of the links, to which they are directly connected
Uses Dijkstra algorithm
Router functions
Routing process:
- Routing protocols
- System config
- Router management
Packet forwarding:
- Header validation
- TTL control
- Checksum check
- Route lookup
- Fragmentation
- Handling IP options
Additionally:
- Packet classification
- NAT
- Traffic prioritization
Router parts
Network interfaces
Forwarding engines
Queue manager
Traffic manager
Backpane
Route control processor
Path Vector protocols principle
Path information is maintained and updated dynamically. Each entry in the routing table
contains the destination network, the next router and the path to reach the destination.
Updates which have looped through the network and returned to the same node are easily detected and discarded.
What is MPLS (Multiprotocol Label Switching)
a forwarding mechanism where packets are forwarded based on labels
MPLS basic functionality
- an analysis of packets entering the network and their classification to FEC classes (Forward Equivalence Class) - recieves label
- labels creation for all the FEC classes
- determination/creation of Label Switched Paths
- labels distribution
- LDP - labels distributed across the network determining the best path to each edge router
- setting the forwarding information tables in the routers
- packets forwarding
- at each inner router label is read and replace with new one
MPLS network components, their functions
Edge Label-Switched Routers (LSRs):
- Ingress LSRs - analyzes packet IP header, assigns FEC label, incorporates the label into MPLS header
- Egress LSRs - removes labels, forward packets to egress link, decrements TTL
Core routers:
- Packet forwarding based on labels
MPLS traffic engineering features
- the ability to establish an LSP that follows a path other than the one
offered as “preferred” by the routing protocol - resources within the network can be dynamically reserved
- traffic can be groomed onto “parallel” LSPs
- traffic can be transferred to
alternate LSPs in the event of a failure - load-sharing and traffic grooming decisions need to be made just once rather than at each node within the network
What is Traffic Engineering
Traffic Engineering is about discovering what other paths and links are available in the network, what the current traffic usage is within the network, and directing traffic to routes other than the shortest so that
optimal use of the resources in the network is made
TCP and its main features
TCP is a connection-oriented service of the transport layer.
Main features:
- Sends acknowledgement of received messages.
- Keeps ordering of packets.
- In case of not receiving a packet, it discards all further packets.
What is congestion control in TCP
an approximate sender’s estimation of available throughput (using cwnd)
TCP Tahoe congestion control mechanism
slow start: exponential increase in cwnd until ssthresh reached.
per RTT without loss:
cwnd = cwnd + 1
per every loss:
ssthresh = 0.5 * cwnd
cwnd = 1
reset to slow start
TCP Reno congestion control mechanism
Receiver sends duplicate ACKs when out of order segments arrived.
With 3 duplicate ACKs loss is indicated (without waiting for timeout):
- performs fast retransmit
ssthresh = 0.5 cwnd
cwnd = ssthresh
- enters fast recovery (without slow start)
TCP Vegas congestion control
Detects congestion based on increasing Round-Trip Time of the packets
in the connection (instead of packet loss like previous two algorithms). When this
occurs, it linearly lowers the cwnd (cwd = cwd - 1).
Why traditional TCP is not optimal in high-capacity networks?
- Limited window size
- Transmission time is small and propagation time is large
- TCP is sensitive to packet losses on high-capacity networks
- Additive increase (for congestion window) is too slow
Source: https://cpham.perso.univ-pau.fr/Paper/TUTORIAL/HOTI-06/HOTI-tutorial-Part2.pdf
Name improved TCP protocols
- Multi-stream TCP
- GridDT
- Scalable TCP
- High-Speed TCP
- BIC-TCP
- CUBIC TCP
- Quickstart
- E-TCP
- fast
- tsunami
Describe Multi-stream TCP
- Multi-stream TCP
Multiple TCP streams transferring one data flow. Improves the TCP’s performance/behavior just in cases of isolated packet losses. Drawbacks: more complicated, startup is accelerated only linearly, overloads caches and buffers.
Describe GridDT TCP variant
- GridDT
A collection of ad-hoc modifications: correction of sstresh (faster slowstart), congestion control modification (cwnd = cwnd + a per RTT, cwnd = b * cwnd per packet loss)
Describe Scalable TCP
- Scalable TCP
Changes the window behavior: cwnd += 0.01 cwnd (per RTT), cwnd = cwnd + 0.01 (per ACK), cwnd = 0.875 * cwnd (per packet loss). For smaller window size and/or higher loss rate in the network the Scalable-TCP switches into AIMD mode.
Describe High-Speed TCP
congestion control AIMD**/MIMD:
cwnd = cwnd + a(cwnd)
. . . per RTT without loss
cwnd = cwnd + a(cwnd)/cwnd
. . . per ACK
cwnd = b(cwnd) cwnd
. . . per packet loss
a and b are functions
** Additive increase/multiplicative decrease
Name two main types of network
Connection-oriented and connectionless.
IPv4 and IPv6 addresses structure
IPv4 32 bit divided into four parts
IPv6 128 bit address in hexadecimal notation divided into 8 groups
CIDR and how it works for IPv4
Classless or Classless Inter-Domain Routing (CIDR) addresses use variable length subnet masking to alter the ratio between the network and host address bits in an IP address.
CIDR notation represents an IP address and a suffix that indicates network identifier bits in a specified format. For example, you could express 192.168.1.0 with a 22-bit network identifier as 192.168.1.0/22.
Source: https://aws.amazon.com/what-is/cidr/
IPv6 addressing (basic structure)
IPv6 address space has 2^128 possible addresses, uses CIDR.
Basic structure:
n bits - global routing prefix (network address)
64-n bits - subnet address
64 bits - interface address
IPv6 MTU path discovery
- a special technique used for determining what size of fragments should be used.
- uses a feedback mechanism performed by ICMPv6’s Packet Too Big messages
What IPv6 uses to find the link address of a node having an IP address?
Instead of ARP protocol as in IPv4, IPv6 uses the Neighbor Discovery Protocol
Neighbor discovery protocol messages types
5 ICMP messages:
*Router Solicitation - (hledame router v siti)
*Router Advertisement - (router se ohlasuje/odpovida na router solicitation)
*Neighbor Solicitation - (hledame sousedy)
*Neighbor Advertisement - (odpoved na hledani sousedu)
*ICMP Redirect - (upozorneni, ze je nalezena kratsi cesta)
Describe IP protocol
The most widespread protocol of the network layer, ensures data (in pieces called datagrams) delivery, even through intermediate nodes (called routers) – host-to-host delivery. Hosts/interfaces are identified by their IP addresses. Provides an unreliable (so-called best-effort) service with connectionless communication. Supplemented by a set of supporting protocols such as ICMP, ARP, RARP, IGMP.
IPv4 and v6 header sizes
IPv4 - 20 to 60 bytes
IPv6 - 40 bytes
TCP handshake
SYN
SYN+ACK
ACK
What is DNS
Bidirectional association of host names to IP addresses (and reverse).
Hierarchical distributed architecture: root servers, authoritative servers, caching
Vulnerabilities: DoS, DNS record spoofing, cache poisoning
Name IPSec framework elements
*AH (Authentication header) - a protocol for authentication
*ESP (Encapsulation Security Payload) - a protocol for encryption
* a definition for the use of cryptographic algorithms for encryption and authentication
* a definition of security policies and security associations between communicating peers
* key management
Security Associations for IPv6 (IPsec)
a set of security information that describes a particular kind of secure connection between one device and another
IPSec modes (names of modes)
Transport and tunnel
K čemu je používán Neighbor Discovery Protocol?
- autokonfiguraci IPv6 adresy (stateful/stateless)
- zjištění prefixu sítě, routerů a jiných konfiguračních informací
- detekci duplikátních IP adres (DAD)
- zjištění L2 adresy uzlu na stejném spojení
- nalezení sousedných routerů pro přeposílání packetů
- udržování si přehledu o dostupných a nedostupných sousedech (NUD)
- detekci změny link-layer adresy
Multicast routing
multicast = delivery of a message/information to a group of destination computers simultaneously in a single transmission from
the source, creating copies automatically in inner network elements (e.g., routers) only when the topology of the network requires it.
employed protocols: IGMP, DVMRP, MOSPF, PIM, etc
