Part I – Internetworking
ISO’s(International
Organization for Standardization) OSI (Open Systems Interconnection) Model:
Layer
|
PDUs
|
Remarks, Examples
|
|
7
|
Application
|
WWW, E-mail gateways,
user interface. Also responsible for understanding the resources needed to
communicate between two devices and establish their availability. SMTP, FTP.
|
|
6
|
Presentation
|
||
5
|
Session
|
Data
|
Keeps different
applications’ data separate, NFS, SQL, RPC, NetBIOS names, X Window. Offers
three modes – full-duplex, half-duplex and simplex. Maintains communication
channels and provides dialogue control. Managing, setting up and tearing down
sessions.
|
4
|
Transport
|
Segments
|
Reliable or
unreliable delivery, error correction before retransmit, TCP/UDP. Performs
flow control, end-to-end connection. Port numbers are used at this layer. Multiplexing,
teardown of virtual circuits. Reassemble the data stream.
|
3
|
Network
|
Data and Route Update
Packets or Datagrams
|
Logical addressing,
routing, IP and IPX. Route update packets are sent at this layer, in addition
to the data packets. Layer 3 devices such as routers break up broadcast
domains and collision domains.
|
2
|
Data-Link
|
Frames
|
Layer 2 devices such
as switches or bridges break up collision domains whereas hubs do not.
Uniquely identifies each device on a local network. This layer uses service
access points, identify network layer protocol used, flow control and
sequencing of control bits (LLC – Logical Link Control - sublayer) and deals
with the protocol access to the physical medium, network topology and error
detection/notification (MAC – Media Access Control - sublayer). A MAC address
on a NIC (Network Interface Card) is a 48 bits address formatted in 12 hexa
digits grouped in twos as such: AF-98-C0-72-A3-2B
|
1
|
Physical
|
Bits
|
Moves bits between
devices, specify voltages, wire speed and pin-out cables. Hubs are also known
as multi-port repeaters and operate at this layer.
|
A layered model enables
different vendors’ products to interoperate (“plug-n-play”), breaks a complex
problem into more manageable entities, eases the changing of one layer without
changing the other. Realize that breaking up a collision (or broadcast) domain
creates more collision domains
Cisco
Hierarchical Model:
Layer
|
Examples
|
Core
|
Large
amounts of traffic reliably and quickly. Fault tolerance important. Don’t use
VLAN, access lists or packet filtering at this layer. Cisco recommends using
layer 2 switches at this layer.
|
Distribution
|
Provides
routing, filtering and WAN access. Place to implement policies on a network
(packet filtering, access lists, queuing, security and network policies,
address translation, firewalls, redistribution between routing protocols,
static routing, routing between VLAN and other workgroup support functions,
definition of broadcast and multicast domains. Cisco recommends using routers
at this layer.
|
Access
|
Continued
access control and policies from the distribution layer, creation of separate
collision domains (and segmentation of contention networks). Cisco recommends
using layer 2 switches at this layer.
|
IEEE
Ethernet (MAC) Standards
|
Ethernet Physical Media
(created by Digital Equipment, Intel and Xerox):
10Base2
|
Thinnet 50-ohm coax
185m, 30 hosts per segment
Physical and logical bus with AUIs
|
100BaseFX
|
Fiber cabling 62.5/125-micron multimode fiber
point-to-point, 400m
ST or SC connectors
|
10Base5
|
Thicknet 50-ohm coax
500m and 208 hosts per segment
Physical and logical bus with AUIs
|
1000BaseCX
|
Copper shielded twisted-pair
25m
|
10BaseT
|
EIA/TIA cat 3,4 or 5, using two-pair unshielded twisted-pair
(UTP) wiring.
100m and 1 user per segment
Physical star and logical bus with RJ-45
|
1000BaseT
|
Cat 5, four-pair UTP wiring, 100m
|
100BaseTX
|
EIA/TIA cat 5,6 or 7 UTP two-pair wiring
100m and 1 user per segment
physical star and logical bus with RJ-45 MII
|
1000BaseSX
1000BaseLX
|
Multi-mode fiber 62.5/50-micron, 260m
Single-mode fiber 9-micron core, 10km
|
Straight-through vs
crossover cables (wire 1<-> wire 3, wire 2<-> wire 6):
Considering the devices
grouped in two categories: 1-switches / hubs / bridges, 2-workstations /
servers / routers
If changing category,
use a straight through cable, else use a crossover cable. (or use
straight-through when one of the port is marked with an X)
Ethernet
Auto-Negotiation: determines the link speed and duplex
status.
Ethernet Frames:
Ethernet II – uses a
two-byte type field instead of the length.
802.3 – cannot identify
the upper-layer protocol
802.2 (802.3 with LLC
information in the data field of the header) – able to identify the upper layer
protocol
SNAP – Subnetwork
Access Protocol – used in AppleTalk and Cisco Discovery Protocol)
Half-Duplex:
contention net using CSMA/CD (Carrier Sense Multiple Access/Collision
Detection) and a backoff algorithm when collision occur.
Full-Duplex:
two communication paths are required and compatible full-duplex NICs. Loopback
and collision detection must be disabled. Sets up a point-to-point connection
with the remote device. There are no collisions on a Full-Duplex link.
Ring LAN:
Token Ring:
standard created by IBM and reflected in IEEE 802.5 with speeds of 4 or 16
Mbps. Stations cannot transmits until they have the token, which they can
reserve using the Reservation Bits.
FDDI (Fiber
Distributed Data Interface): token-passing ANSI standard providing LAN speed of
up to 200 Mbps if dual rings are active. Only LAN topology that is both
physical and logical ring.
MSAU:
MultiStation Access Unit, the controller of the token ring LAN, for up to 8
stations.
NAUN:
Nearsest Active Upstream Neighbour
Active Monitor:
one station on the ring always ensures there is only ever one token on the
ring.
Beaconing:
process by which a station attempts to determine a network failure.
No comments:
Post a Comment