Great answer, Howard. We like to think of a router as having a simple table like this:
prefix next-hop prefix next-hop prefix next-hop But, from my limited knowledge, I don't believe any modern day router really stores "address lookup" data in this way. They don't use trees any more either. In recent reading, I found information about multbit tries (not a typo). I was just surfing, but I think you can find information about tries at Cisco's site too. CEF uses a trie also. Also, from my surfing I have discovered that computer science departments these days have computer networking classes in "Switching Systems." These classes cover the processing required of any network switch, whether it's switching cells, L2 frames, L3 packets, etc. This course at Washington University in St. Louis has a good Web site: http://boushi.arl.wustl.edu/~jst/cs/577/ Well, it is a Friday night in March. So instead of looking at more university computer science Web sites, I think I will now go watch some college basketball! Priscilla At 07:50 PM 3/15/02, Howard C. Berkowitz wrote: >At 6:44 PM -0500 3/15/02, mlh wrote: > >no, IRB is a technique integrating routing and bridging. > >but switching routers use hardware switching technique > >which based on layer 3 address not layer 2 address. > >this is one of the differences between router switches > >and switching routers. > >it's my point, pls correct it. > > >Again speaking from router design experience, the difference between >layer 3, label, and layer 2 forwarding is less profound than one >might think. > >The frame arrives on the ingress port and gets buffered and >validated. I'll ignore VLANs. Assuming it were an Ethernet frame, the >forwarder would extract the first 6 bytes. If it were an ordinary >packet, it would extract the destination IP address (Bytes 12-15 >following the LLC field if one is present). If it's MPLS, it extracts >the 32 bits following the data link header. > >MAC addresses get looked up directly, possibly in a CAM. Admittedly, >that's probably faster than radix tree lookup, discussed next, but >destination lookup time is rarely the limiting factor in >router/switch performance. It's pretty much a solved problem. The >size of routing/forwarding tables is much more limited by the ability >of the routing protocols to reconverge in a reasonable time than the >time of search. FIBs generally have a constant lookuo time. > >Some masking is applied to the IP and MPLS addresses, and then the >prefix or label is usually looked up in some sort of radix tree in >static RAM. > >Any of these searches pop up the destination port (not address) if >one is known, plus other information about processing (e.g., QoS). >Statistics are incremented, error check fields are checked, and the >packet/frame goes to the fabric. Crossbar fabrics usually are >nonblocking, while shared busses may need buffering. > >Bottom line: layer 2 switch, layer 3 switch, MPLS switch, router, >etc., are more marketing than architectural distinctions. There are >differences between MPLS Label Edge Routers and Label Switched >Routers, before we get into the GMPLS non-packet forwarders (e.g., >wavelength switches). > >Remember what Cisco calls something in its literature, but don't >obsess over arbitrarily fitting one box into a category. > > > > >----- Original Message ----- > >From: "Chuck" > >To: > >Sent: Friday, March 15, 2002 11:24 AM > >Subject: Re: Layer 3 switch ? [7:38358] > > > > > >> so if I enable IRB on my 2501, I now have a layer 3 switch? ;-> > >> > >> > >> ""mlh"" wrote in message > >> [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > >> > According to Clark's Cisco LAN Switching(page 452 ), > >> > layer 3 switching techniques can be grouped > >> > into two categories: > >> > Routing switches > >> > Switching routers > >> > > >> > mlh > >> > > >> > ----- Original Message ----- > >> > From: "Larry Letterman" > >> > To: > >> > Sent: Friday, March 15, 2002 2:48 AM > >> > Subject: RE: Layer 3 switch ? [7:38358] > >> > > >> > > >> > > A true router, 2621/3640/7200, is not usually considered > >> > > a L3 switch. A 6500 with an MSFC module installed can be > >> > > is a L3 switch and will perform L2/L3 routing and switching. > >> > > > >> > > A 6500 without the MSFC module is just a large high speed switch, > >> > > capable of only L2 switching. > >> > > > >> > > A layer 3 switch usually routes the first packet in the flow of data > >> > > and then switches the rest in the switching hardware. This is why L3 > >> > > routing/switching is quite a bit faster. A traditional router will >use > >> > > IOS software to determine routes and the switch each packet between > >the > >> > > interfaces in the router. > >> > > > >> > > > >> > > Larry Letterman > >> > > Cisco Systems > >> > > [EMAIL PROTECTED] > >> > > > >> > > > >> > > -----Original Message----- > >> > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf >Of > >> > > John Green > >> > > Sent: Thursday, March 14, 2002 10:50 PM > >> > > To: [EMAIL PROTECTED] > > > > > Subject: Layer 3 switch ? [7:38358] > >> > > > >> > > > >> > > Is it ok to refer to a "router" as a Layer 3 switch ? > >> > > > >> > > cisco 6500 was referred to as a Layer 3 switch. > >> > > > >> > > question: does it(6500) have routing capabilities ? > >> > > ----------------------------------------------------- > >> > > > >> > > to connect to different vlans one needs a router. > >> > > right ?? (as shown below) > >> > > switchA --------ROUTER-------switchB > >> > > > >> > > but say some nodes connected to switchB are on the > >> > > vlan of switchA. so now to connect switchA and switchB > > > > > can router be ok ? > >-- >"What Problem are you trying to solve?" >***send Cisco questions to the list, so all can benefit -- not >directly to me*** >******************************************************************************** >Howard C. Berkowitz [EMAIL PROTECTED] >Chief Technology Officer, GettLab/Gett Communications http://www.gettlabs.com >Technical Director, CertificationZone.com http://www.certificationzone.com >"retired" Certified Cisco Systems Instructor (CID) #93005 ________________________ Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=38515&t=38358 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
