Docstoc

Quick Guide - Layer 2 Switching

Document Sample
Quick Guide - Layer 2 Switching Powered By Docstoc
					CCNA4.com

Quick Guide - Layer 2 Switching
LAYER 2 SWITCHING - Switching breaks up a large collision domain into smaller ones. Switches operate at the Data Link Layer via the use of MAC addresses. - Spanning Tree Protocol -- eliminates loops in a Layer-2 switched network. - Collapsed backbone (pre-switch design) -- all hosts needed to go via the backbone to reach any network services. - Switches use ASICs (Application Specific Integrated Circuits) to build and maintain their filter tables. - Switches are faster than routers because they only look at MAC addresses. There is no modification to the data packet. - Bridges are software based, while switches are hardware based. Bridges can only have one Spanning Tree instance per bridge, switches can have many. Bridges can only have up to 16 ports. - Important Layer-2 switch functions: - Address learning -- the source MAC address of each received frame is placed in a forward/filter table. - Forward/filter decisions -- after the destination MAC address is found in the table, a packet is sent through the appropriate exit interface. - Loop avoidance -- with the help of STP (Spanning Tree Protocol). - If no loop avoidance scheme is in place, broadcast storms are possible. - Thrashing -- when a switch is so busy updating the MAC table that is fails to forward a frame. - Spanning Tree Protocol -- DEC (now known as Compaq) created the original version of STP. Later, IEEE introduced 802.1d, which is what all Cisco switches use. 802.1d is not compatible with the DEC version. - STP uses the spanning-tree algorithm (STA) to create a topology database, and then find and destroy redundant links. - Spanning Tree terms: - Spanning Tree Protocol -- a bridge protocol that uses the STA to find redundant links and create a topology database. - Root Bridge -- this is the bridge with the lowest Bridge ID -- all decisions are made from the perspective of this bridge. - Bridge Protocol Data Unit (BPDU) -- switches exchange control information in BPDUs. - Bridge ID -- 8 bytes long -- a combination of the bridge priority (32768 by default on all switches) and the base MAC addresses. The lowest Bridge ID becomes the root bridge. - Non-root Bridge. - Root port -- a link directly connected to the root bridge, or the shortest path to the root bridge. If more than 1 port, bandwidth is checked. The lowest cost port becomes a root port. - Designated port -- a forwarding port (can be a root port too) -- if a bridge has 2 ports and it is not a root bridge, one of the ports is root, and the other is designated (forwarding). - Port Cost -- determined by the bandwidth of a link. - Non-designated port -- can be blocking or forwarding. - Forwarding port. - Blocked port -- a port that will not forward frames in order to prevent loops. - BPDUs are sent every 2 seconds. - Calculating STP cost -- new and original values: - 10Gbps -- new cost 2, originally 1. - 1Gbps -- new cost 4, originally 1. - 100Mbps -- new cost 19, originally 10. CCNA4.com

CCNA4.com - 10Mbps -- new cost 100, originally 100. - The 1900 series switches still use the original cost specifications. - Spanning Tree Port States: - Blocking -- a blocked port does not forward frames. It just listens to BPDUs. - Listening -- the port listens to BPDUs. - Learning -- the port learns MAC addresses and builds a filter table, but does not forward frames. - Forwarding -- the port sends and receives all data. - Switch ports are most often in either blocking or forwarding state, unless a change to the network has just been made. - Convergence -- the time it takes for a port to switch from blocking to forwarding mode, or vice versa -- usually 50 seconds. No data is forwarded during this time. It is NOT recommended to change the default STP timers. - A root bridge ALWAYS has every port in forwarding mode. - Switching Modes: - Cut-Through -- the switch looks up the destination MAC address in the MAC filter table and starts forwarding the frame immediately -- before the frame is over -- 13 bytes into the frame. This method is fastest, but has no error checking. - FragmentFree (Modified Cut-Through) -- the switch reads 64 bytes from the frame (waits for the collision window to pass) and then forwards the frame. This is the default method for 1900 series switches. The first 64 bytes is where almost all errors happen. - Store-and-Forward -- a complete frame is received, a CRC is run (invalid CRC is either <> 1518 bytes). This is the slowest method, but with the most error checking. Catalyst 5000 switches use this method, and the setting CANNOT be modified. By badboy

CCNA4.com


				
DOCUMENT INFO
Description: CCNA Exam, Exam Answers, the network, ccna test, Module 2, Module 8, Certification Exam Preparation, Free Resources, Packet Tracer, cisco ccna training, Module 3, IP address, Module 4, Module 5, subnet mask, Quick Notes, IP address, in the Frame, encapsulation type, Frame Relay header, the Frame, Inverse ARP, Relay switch, error checking, Associate Exam, Leading the way, Best Practice, CCNA Exam, virtual circuits, Cisco Certified Network Associate, IP access, filter packets, Free Cisco,