If you can believe the CIT class, the small, middle, big, very big, large,
and huge buffers are used for process switching. These are the buffers that
you track with the "show buffers" command. These buffers are in system
memory and they generally get used if you have disabled fast switching and
other performance-enhancing forwarding methods.
System buffers are also used for packets that the router itself generates,
such as ICMP packets, routing protocol updates, SNMP, and Novell IPX SAP
updates.
With process switching, when a packet arrives at an interface, the
processor is interrupted for the time it takes to copy the packet from the
interface buffer to system memory. The processor looks up the Layer-3
destination address in the routing table to determine the exit interface.
The packet is rewritten with the correct header for that interface and
forwarded to the interface. At this time, an entry is also placed in the
fast-switching cache so that subsequent packets for the destination address
can use the same header.
With fast-switching, a packet is handled immediately, without scheduling an
interrupt of the system processor and without using those small, medium,
big, etc. system buffers.
Having different size buffers means that memory is used efficiently.
Packets come in different sizes. Why waste a 1500-byte buffer for Telnet
packets, which are only 64 bytes, typically? Or another example is
AppleTalk packets, which are always 600 bytes or less. IPX used to use
600-byte packets, though Large IPX has been around for a while. IPX SAP
packets are still only about 600 bytes if I remember correctly. Many
distance-vector routing protocols also use packets in the 512-600 byte
ranges for their routing updates. Database applications use medium-sized
packets usually. HTTP also often uses medium-sized packets, especially if
the browser uses multiple TCP sessions to download information.
If you are process-switching and there are lots of dropped packets, you can
issue a "show buffers" command and determine which size packets are causing
the problems. Knowing the size of the packets that are getting dropped
helps you identify which protocols are the culprits.
Priscilla
At 11:36 PM 12/12/00, CCIE TB wrote:
>Hi,
>
>There are many buffers on Cisco routers like small, middles, big, ....
>1. Is there any advantage/disadvantage of having many buffers compared to
>having one or two only.
>2. How and in what order they are being used by the system ?
>
>Any feedback is highly appreciated.
>
>Adia
>
>_____________________________________________________________________________________
>Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
>
>_________________________________
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
