Hej, Am 11.02.2009 19:13 Uhr, John Osmon schrieb: > On Wed, Feb 11, 2009 at 04:11:38PM +0100, Mathias Wolkert wrote: >> I like the idea of having one physical version showing cables and devices >> (CDP/EDP/LLDP view pretty much) and one logical view showing IP subnets. >> Many times I found *documented* networks where this is all combined making >> it very unclear. >> The hard part is to visually show what VLANs are active in each switch.
> Most networks need at least two diagrams: > - a logical map showing network boundaries/collision domains/etc. > (This is where VLANs get documented) > - a physical map showing *how* things are connected. > (This is where equipment and their interconnects are documented) the actual needs strongly depend on the design of the network. If your network is segemented by many routers, it may even be sufficient to do a dozen or so traceroutes and parse the results ;-) If you run flat, switched networks with hundreds of switches but only few routers and possibly extreme heterogeneous subnetting in a multi-vendor environment, you do not get very far by parsing configs or "autodiscovering" the net. It becomes even more interesting if you run active layer 1 equipment like DWDM boxes or radio connections :-) Personally, I think most important is a clean documentation of Layers 1 and 2 AND the corresponding contact data for 3rd party sites/lines/equipment. These are the things you cannot get easily out of your network, and when experiencing failure on that level, you'll be happy to have this information on one single map. Always remember: Layer 3 is easy. Routing is easy. You have a lot of tools and deterministic protocols. Layers 1/2 are the wild jungle where you may see strange things happen and are partly blind and constrained. Combining the maps for Layers 1 and 2, by the way, is possible. Use colours, line types, and again geometric figures. Regards, .m
signature.asc
Description: OpenPGP digital signature
