Connecting LANs: Section 15.1

29
Connecting LANs: Section 15.1

description

Connecting LANs: Section 15.1. Figure 15.1 Five categories of connecting devices. Layer 1 (repeaters, hubs). device simply regenerates signals defines a collision domain Everything connected by a layer 1 device lies in the same collision domain. Dumb - PowerPoint PPT Presentation

Transcript of Connecting LANs: Section 15.1

Page 1: Connecting LANs:  Section 15.1

Connecting LANs: Section 15.1

Page 2: Connecting LANs:  Section 15.1

15.2

Figure 15.1 Five categories of connecting devices

Page 3: Connecting LANs:  Section 15.1

Layer 1 (repeaters, hubs) device simply regenerates signals defines a collision domain

Everything connected by a layer 1 device lies in the same collision domain.

Dumb Not as common as they once were since more

intelligent devices are affordable

Page 4: Connecting LANs:  Section 15.1

15.4

Figure 15.2 A repeater connecting two segments of a LAN

Page 5: Connecting LANs:  Section 15.1

15.5

Figure 15.4 A hierarchy of hubs

Page 6: Connecting LANs:  Section 15.1

Layer 2 (bridges, switches) Routes based on physical (MAC) addresses. Implements contention protocol. defines a broadcast domain. Routing tables used to determine where to forward

frames.

Page 7: Connecting LANs:  Section 15.1

15.7

Figure 15.5 A bridge connecting two LANs

Page 8: Connecting LANs:  Section 15.1

Routing logic Look for the destination MAC address in a routing

table. If found, send frame out over the specified port. If not found, send frame out over ALL ports except

the one from which it arrived (flooding).

Page 9: Connecting LANs:  Section 15.1

How do routing tables get built? Manually (static)

Only the earliest bridges Learning

Examine the source MAC address of every frame. Store that address and the port over which the frame

arrived into the routing table. Every so often purge the routing tables and repeat in

case devices move

Page 10: Connecting LANs:  Section 15.1

15.10

Figure 15.6 A learning bridge and the process of learning

Page 11: Connecting LANs:  Section 15.1

Routing Logic for a bridge/switch Look up destination address in routing table. If the address corresponds to the port over which it

arrived, ignore it. If the address corresponds to a different port, send

the frame out that port. If the address does not exist in the routing table, send

it out all ports (flooding) except the one over which it arrived.

In any case, store the frame’s source address in the routing table along with the port over which it arrived.

Page 12: Connecting LANs:  Section 15.1

Should avoid loops in the LAN topology

Page 13: Connecting LANs:  Section 15.1

15.13

Figure 15.7 Loop problem in a learning bridge

Page 14: Connecting LANs:  Section 15.1

A system admin might want redundant bridges as backups if one fails.

Redundant bridges means multiple routes and, hence, loops.

IEEE created a standard by which bridges can execute an algorithm that creates a spanning tree a minimum set of connection among all LANs

IEEE 802.1d Cisco

Page 15: Connecting LANs:  Section 15.1

It’s a clever algorithm in which bridges flood the network, sending information about themselves and the LANs to which they connect.

No time to cover this, so you can skip it. Also, lifetime of such devices is usually measured in

years so failures are rare.

Page 16: Connecting LANs:  Section 15.1

Bridging different LANs (legacy) Change frame formats. Sometime remove a priority field (losing it) or

adding a default priority. If frame was too large to route onto another LAN

Further limit frame size at originating device beyond the normal protocol limit – loses transparency.

Divide frame and reassemble (router function). Sometimes the term brouter (bridge/router) was used.

Page 17: Connecting LANs:  Section 15.1

Layer 3 (routers) Routing tables based on IP addresses Slower that switches since there are more layers to

process Unlike switches, can be connected in loops have logic to detect recirculating packets For example, an Internet packet has a TTL (time-to-

live) field that decrements each time it passes through a router.

If it decrements to 0, the router drops it.

Page 18: Connecting LANs:  Section 15.1

Backbone networks Section 15.2 Not much in this section That part of the network that connects other

components of the network

Page 19: Connecting LANs:  Section 15.1

15.19

Figure 15.12 Bus backbone

Page 20: Connecting LANs:  Section 15.1

VLANS Section 15.3 Nice reference at WUSTL Historically, LAN constrained by physical location

and a LAN defined a broadcast domain. LANs were separated for better security and

efficiency.

Page 21: Connecting LANs:  Section 15.1

15.21

Figure 15.15 A switch connecting three LANs

Page 22: Connecting LANs:  Section 15.1

A VLAN is a collection of devices that behave as though they were connected to the same LAN

Ethernet frame extended to include a VLAN ID. Above layer 2, but below layer 3.

Bridge port usually determines which VLAN, but MAC addresses and protocols used can be the deciding factor.

Page 23: Connecting LANs:  Section 15.1

15.23

Figure 15.16 A switch using VLAN software

Page 24: Connecting LANs:  Section 15.1

Some advantages smaller broadcast domain issues in one VLAN don’t affect another Another layer of security limits “reach” of devices within layer 2 (going

beyond the reach requires layer 3 device where administration can enforce more policies).

Page 25: Connecting LANs:  Section 15.1

May help to contain viruses since many are spread by broadcast traffic.

Campus uses up to 35 VLANs in res-life (or, at least, used to)

Touring companies for Weidner activities. Their PCs are often infected and a VLAN helps isolate the problems.

Page 26: Connecting LANs:  Section 15.1

Spanning Tree Algorithm Outline (SKIP) Assign a cost to a bridge port. (might be inversely

proportional to bit rate, reflect queue sizes or other bridge design factors, or just =1)

Cost of sending a message = sum of costs of bridge ports (sending only)

Visualize LAN topology as a graph

Page 27: Connecting LANs:  Section 15.1

Bridges elect a root bridge. Each bridge sends BPDUs (contains Bridge ID, port ID over which BPDU first sent, accumulated cost of ports over which BPDU is received-this is cost back to sender)

Bridge receives BPDU: If own ID is larger than that in BPDU, stop sending its own BPDUs, update received one, and forward on all other out ports. If own ID smaller or equal, then do not forward. Packets could circulate endlessly, so drop if cost > threshhold value.

Page 28: Connecting LANs:  Section 15.1

After a time, only one bridge is sending BPDUs and all others receive BPDUs from only 1 source. All know identity of root bridge and cost to get there.

Determine designated bridge for each LAN (bridge to forward frames from each LAN). Bridges send other BPDUs over each LAN to which they are connected (not one connected via root port). Each BPDU contains cost to root bridge.. Bridge with cheapest cost to root bridge is the designated bridge for that LAN.

Page 29: Connecting LANs:  Section 15.1

Can use timers to periodically reconfigure. Less practical than they once were since switches are

so reliable. Also backup will have been idle for a very long time and connections can go bad.

Time to execute sp tree algorithms causes timeouts More useful if there are multiple switches in a

hierarchy.