Standard Ethernet Detection for Packet Errors on the Wire
Advanced System Diagnostics and Troubleshooting Guide
27
Standard Ethernet Detection for Packet Errors on
the Wire
Data transiting from one switch to another is checked for packet errors using the Ethernet Cyclic
Redundancy Check (CRC) built into the IEEE 802.3 specification.
As the sending switch assembles a frame, it performs a CRC calculation on the bits in that frame and
stores the results of that calculation in the frame check sequence field of the frame. At the receiving end,
the switch performs an identical CRC calculation and compares the result to the value stored in the
frame check sequence field of the frame. If the two values do not match, the receiving switch assumes
that packet data has been illegally modified between CRC calculation and CRC validation and discards
the packet, and increments the CRC error counter in the MAC device associated with that port. In
Extreme Networks devices, ExtremeWare polls the MAC CRC error count registers and makes that
information available through the output of the
show port rxerrors
CLI command.
Extreme Networks’ Complementary Detection of Packet
Errors Between Wires
The 802.3 Ethernet specification provides a CRC check for validation of data on the wire, but offers no
guidance for handling data validation in the devices between the wires. Because these devices are far
more complicated than the wires connected to them, common sense indicates the requirement for some
mechanism for checking internal data integrity. To complement the Ethernet CRC data validation
scheme, Extreme Networks switches implement an internal data checksum validation scheme referred
to as the
fabric
checksum
.
The switch fabric in a switch is essentially an extremely high-speed data path connecting multiple ports
and using a set of programmable lookup tables to make intelligent forwarding decisions when moving
data from point to point inside the switch.
Figure 8
uses a generalized block diagram of a switch to
illustrate data movement within a switch.
Fast path
This term refers to the data path for a packet that traverses a switch and does not
require processing by the CPU. Fast path packets are handled entirely by ASICs
and are forwarded at wire rate.
Slow path
This term refers to the data path for packets that must be processed by the switch
CPU, whether they are generated by the CPU, removed from the network by the
CPU, or simply forwarded by the CPU.
Table 3: Data Error Terms (continued)
Term
Description
Содержание ExtremeWare Version 7.8
Страница 8: ...8 Advanced System Diagnostics and Troubleshooting Guide Contents...
Страница 14: ...14 Advanced System Diagnostics and Troubleshooting Guide Introduction...
Страница 24: ...24 Advanced System Diagnostics and Troubleshooting Guide i Series Switch Hardware Architecture...
Страница 48: ...48 Advanced System Diagnostics and Troubleshooting Guide Software Exception Handling...
Страница 102: ...102 Advanced System Diagnostics and Troubleshooting Guide Additional Diagnostics Tools...
Страница 110: ...110 Advanced System Diagnostics and Troubleshooting Guide Troubleshooting Guidelines...
Страница 114: ...114 Advanced System Diagnostics and Troubleshooting Guide Limited Operation Mode and Minimal Operation Mode...
Страница 120: ...120 Advanced System Diagnostics and Troubleshooting Guide Index...