4 - Internet Architectures Flashcards
Computer Networks
A computer network is a collection of computers and devices connected for the purpose of electronic data
communication that allows them to share information and services
Private Networks
requires users to obtain permission to gain access; Either manually by a network administrator or via password
Public Networks
access is not restricted (e.g., the internet)
World Wide Web
information space (on the Internet) in which global identifiers called Uniform Resource Identifiers (URI) identify the items of interest, referred to as resources
WAN
Wide Area Network:
Connects computing devices over a large geographical distance or even those across the globe
MAN
Metropolitan Area Network:
Interconnects computing devices within a city or a metropolitan area
LAN
Local Area Network:
Connects computing devices within the same room or building
URI
A Uniform Resource Identifier (URI) identifies a resource either by location, or a name, or both
URL
Uniform Resource Locator
- Is a URI that identifies a resource and also provides the means of location of the resource by describing the way to access it
URN
Uniform Resource Name
- Is a URI that includes a name within a given space , but does not describe how to access the resource
Tier 1 ISPs
Are operated by national telecommunication
companies (e.g., Deutsche Telekom, AT&T);
Tier 1 networks exchange data directly with each other
Tier 2 ISPs
- They exchange Internet traffic through peering agreements and purchase Internet traffic from Tier 1 ISPs
- Data exchange is done at neutral data centers, that is,
data centers in shared use by multiple ISPs, called
Internet Exchange Points (IXPs) or MXP
Tier 3 ISPs
- Delivers Internet access to residential homes and businesses
- Strictly purchases Internet traffic from higher tier networks
POP
Point of Presence
- Local access point of an ISP where the telecommunication lines from commercial or domestic buildings are connected to the ISP’s network
- Often located within the facility of a telecommunications
provider responsible for the infrastructure to the customer
Internet Society (ISOC)
Provides organizational structure to support the process of Internet standard development
Internet Engineering Task Force (IETF)
Develops and maintains voluntary Internet standards (e.g., TCP/IP)
Internet Corporation for Assigned Names and Numbers (ICANN)
Responsible for the IP address space allocation and management of the domain name system
Word Wide Web Consortium (W3C)
- Responsible for developing interoperable technologies for the WWW
- Examples: HTML, XML, CSS, and SOAP
Internet Protocol Suite (TCP/IP)
The Internet protocol suite is a set of protocols that enables Internet communication by specifying data transmission, addressing, and routing
- defines how data should be packaged, addressed, sent, routed, and received
Application Layer
provides applications with standardized interfaces that allow them to send data to other applications or receive data from them via a network
Transport Layer
responsible for the correct transfer of data between network nodes, independent of their application, specific data structures, and underlying network
- TCP ensures that, for instance, data arrive in the order they were sent, that they arrive correctly, that duplicate data are discarded, and that data lost in transit are resent
UDP
provides highly efficient but less reliable data transmission and has no error-recovery mechanism
Transmission Control Protocol (TCP)
- TCP is a reliable service which guarantees that all bytes are received in the right order
- This is done by using positive acknowledgements (ACK) with re transmission; receiver responds with an ACK for ever data packet he receives
- Sender retransmits packets for missing ACKs after a given time
Network Layer
responsible for transporting data between the right nodes within a network or across multiple networks; node addressing and data routing
Data-Link Layer
provides an interface to the actual physical networking hardware that links two or more distributed nodes
IP Addresses
An IP address is a unique string of numbers separated by full stops that identifies each computer using the Internet Protocol to communicate over a network.
Domain Name System (DNS)
The Domain Name System (DNS) is a hierarchically structured, distributed set of databases that maps IP
addresses to corresponding domain names
Internet Protocol (IP)
IP is responsible for addressing host interfaces, encapsulating data into datagrams and routing data from a source host to a destination host
IPv4
- 32 bit long binary string
- Specifies 232(4,294,967,296) unique addresses
- Pool of addresses became to small due to rapid growth of the Internet
IPv6
- 128 bit long binary string
- Specifies 2128(approx. 3.403 x 1038) unique addresses
Packet Switching
Packet Switching describes a switching and transmission technology which splits complete messages into smaller packets. These packets can be transmitted along different lines of a network and they are re assembled into the original message by the receiving host.
+ channel capacity
+ transmission latency
+ robustness
IP Packets: Datagrams
- Header: keeps information about the packet, the service, and other transmission related data (e.g., IP address of sender and destination, sequence number of the packets)
- Payload: holds the actual carried data
- Each layer treats information from above layers as data
- the process of preserving the data while attaching a new header is known as encapsulation
Router
On a network, a router is a device that determines the best path for forwarding a data packet toward its destination. A router is connected to at least two networks and is located at the gateway where one network meets another
- Routers check whether incoming packets contain errors
Switches
interconnect nodes within the same network
Maximum Transmission Unit (MTU)
maximum data size of a packet sent or received in a single network transaction
Domain Name
A domain name consists of one or more labels, separated through decimal points
- Each label specifies a subdomain of the domain
- Subdomains are organized hierarchically in a tree like structure, starting from the nameless DNS root domain
Top level domains
- Generic top level domains: com, info, net and org
- Country code top level domains: de, fr , ca,
- Sponsored top level domains: edu , gov , jobs
2nd level domain
Open for reservation by organizations and end users (e.g ., KIT, google, Facebook)
Sub domain / 3rd level domain
- not mandatory
- Used to specify a certain server inside an organization
(e. g.: www[.examaple.com], mail[.google.de], studium[.kit.edu]
Content Delivery Networks (CDNs)
are a collection of network devices that are controlled by a common management infrastructure with the main purpose of delivering content (e.g., websites, videos) more effectively to clients over the Internet.
Funktionsweise CDN
- move content to network location closer to consumer
- Reduction of distance decreases latencies, risk of connection interruptions, improves transmission speed
- replicating the content onto multiple content delivery servers in different geographical locations (surrogate servers)
- Clients’ content requests are automatically routed to the surrogate servers
- CDNs are often operated by specialized network service providers such as Amazon Web Services (AWS)
Steps DNS Lookup
Example domain: aifb.kit.edu/news.html
1) User wants to visit a website using its domain name, so the browser sends a request to a known DNS server
2.8) In case the domain name is known to the server, as it is stored in the cache, it will directly provide the relevant DNS record
3) The right server is contacted
4) User receives the requested “/news.html”
or
2.2) If a DNS server does not know the requested domain name, it will refer to the DNS server for the root level domain
2.3-6) The root level DNS server refers to top level DNS servers, which in turn refer to 2nd level DNS servers until the IP is known
or
2.7) The DNS server on the 2nd Level provides the IP address of the requested domain name
3) The right server is contacted
4) User receives the requested ”/news.html”
content delivery infrastructure
When a client requests a single content item from a CDN, the request is directed to the best suited surrogate server with a copy of the item
request-routing infrastructure
directing a content request from a client to the right surrogate server (shortest delivery time is not necessarily only dependent on server’s geographic location)
distribution infrastructure
All content is first published as a master copy on an origin server. Master copy is then replicated on one or more surrogate servers. Changes are first applied to the
master copy on the relevant origin server and afterwards
replicated on the surrogate servers
accounting infrastructure
measuring and recording the networks’ content distribution and delivery activities
Private CDN
A globally operating cooperation can build their own CDN to deliver content to their different
subsidiaries across the globe
Federated CDN
- Based on infrastructure that is operated by multiple content or service providers
- Participating providers pool their existing resources into a single delivery network
Two possible deployment approaches:
- Bilateral approach: every participating provider directly interconnects with every other provider in the network
Exchange approach: every participating provider connects to a central hub that provides internetworking functionalities (e.g., routing)
Peer to peer (P2P) CDN
- Surrogate servers are either partly or completely substituted by the network‘s clients (i.e peers), which both provide and consume the content
- Creates a mesh network consisting of users who want to access the same content
- Coordinates its clients so that they send chunks of the item to each other
Software-Defined Networking (SDN)
Software Defined Networking (SDN) is an emerging network architecture in which network control is
decoupled from the forwarding devices and is directly programmable
Objectives of SDN
- prioritize one customer’s traffic over another
- Consolidate control system that instructs the network
devices on how to forward packets and which can be
centrally administered
SDN controller
control hub enabling applications from
the layer above to dynamically change how the individual
network devices on the data layer below handle the
network traffic
Southbound APIs
connect the control and infrastructure
layer. Sends configuration and routing information to the
physical network devices
Northbound APIs
enable communication between the SDN controller and the software applications above that require network services to fulfill their intended function
Network devices
transmit operational data required for monitoring indented the network traffic up to the control layer
Dis-/Advantage of SDN
+ loose coupling between the three layers enables virtualization and the dynamic allocation of network and
service functions
+ increased scalability
+ ability to monitor and adapt network resources quickly
+ detect network anomalies caused by malicious behavior
- introducing a single point of failure into the network, which may decrease its reliability
Overlay Networks (SDN overlays)
An overlay network is a virtual network of nodes and logical links built on top of an existing network in
order to implement a network service not available in the existing network
Overlay Networks (SDN overlays): Funktionsweise
- create an additional software layer that abstracts an underlying network’s functionality
- software layer can then be customized to provide specialized services
- many well-known Internet services can be classified as overlay networks running on top of the public Internet (virtual private networks (VPN), peer-to-peer file sharing, and voice over IP (Skype))
- Internet itself was an overlay network. Internet started as a network interconnecting computer nodes via preexisting public telephone networks; the Internet overlaid the physical telephone infrastructure
tunnel protocols
encapsulate traffic inside IP packets, decoupling this traffic from the standardized TCP/IP processing logic, which in turn creates a virtual tunnel running through the
network
Dis-/Advantages Overlay Networks
+ useful when the underlying network cannot be easily
customized
- increasing the performance overhead costs and the communication complexity
- more difficult to track down the root causes of performance or availability issues