Hi Eunseok,
There may be technical and administrative reasons for that.
As just an example, the technical reasons may include using a non-routable protocol on stations connected to different switches that nonetheless need to talk to each other using that protocol. An example can be the NetBEUI protocol for file and printer sharing in old Windows network. Another reason may be that, say, the stations are using multicast but the network itself is not enabled for multicast routing, so if the stations were placed in different VLANs, they would not hear each other. Yet another need for the same VLAN is simply the management VLAN that is probably the same for all manageable switches in a common building switch block (I surely wouldn't be creating 50 unique VLANs for 50 access switches, one for each - maybe I am obsolete here :) ).
Administrative reasons can cover anything that is not that technical related. Perhaps the migration is too costly and time consuming, and you cannot afford to have such a long downtime. Perhaps everyone is so much used to the existing design which works fine for them that moving off to a different design would be a useless exercise. Perhaps the switches do not even support as many VLANs as would be required by the new design (although this is technical, yeah - but it's a reason alright).
Many similar reasons could be invented - but I guess you get the picture.
Anyway: What's the book you were reading? A VLAN prevented from spanning multiple switches? That's quite a requirement the book proposes. I know that VLANs do not span the distribution layer, but now they want it not to even span multiple access switches under the same distribution layer?
Best regards,
Peter
