From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Fri Jun 04 2004 - 02:18:20 GMT-3
At 10:38 PM -0400 6/3/04, asadovnikov wrote:
>I have to say that theory of VTP V3 certainly looks more appealing to me
>that previous versions. It is however fairly new and I trust requires CatOS
>8.1 (VLAN information only) or 8.3 (VLAN + MST), and I do not believe it is
>in IOS yet. As the software features are added and the software which
>supports it becomes general deployment, V3 going to become an important
>player. It is hard for me to predict though how important as timeline and
>scope of support hard to predict at this point.
>
>Having said all that the reason I do not like VTP has little to do with
>features of a particular version of a protocol; I rather do not see a useful
>place for VTP from design perspective.
Given your assumptions about L3 being better in most cases, I agree
completely with what you are saying, especially in an enterprise
context. There is some role for VLANs as a multiplexing protocol for
Metro Ethernet service provider applications, but these are either
customer-to-POP or customer-to-concentrator-to-POP. GMPLS strategies
may yet replace some of those applications, but not overnight.
>
>Back when flat earth was a way to design networks VTP was useful. This days
>the networking trend way I see it is to run IP routing on as many boxes as
>possible. Advantages of running L3 box versus to L2 box are numerous; and
>the only disadvantage is price. The dollar premium paid for L3 box was
>coming down every year and I trust has reached the point when benefits of
>extra functionality overweight few extra dollars in cost. Going forward I
>would expect this trend to continue, price gap to become smaller, extra
>features of L3 boxes be more important.
>
>I could go on for hours on how L3 technologies are better understood, better
>scale, better work... To cut the story short... Certainly on the
>distribution and core I would expect this days everybody to run L3; and I do
>not expect VLANs to span campuses any longer, I expect VLANs to be used on
>trunks between boxes with each vlan be local to 1-2 (3 on rare occasions)
>boxes.
Or many-to-one or many-to-two in SP applications, using Ethernet over
fiber or CATV.
>Given this I do not see how VTP makes anything any better, there is
>simply no need for VTP to assist when scope is so small, but it can be
>involved upkeep even on such small scale.
Very reasonable. Also, as I mentioned, it's less important when
we've converged on one VLAN/ELAN strategy rather than several.
>
>Best regards,
>Alexei
>
>-----Original Message-----
>From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On Behalf Of
>Howard C. Berkowitz
>Sent: Thursday, June 03, 2004 9:25 PM
>To: ccielab@groupstudy.com
>Subject: RE: Connecting two core switches / Design
>
>
>At 7:46 PM -0400 6/3/04, asadovnikov wrote:
>>Assuming that your core is L3 (which it really should be), for 2 gig I
>>would rather see you use 2 independent point-to-point L3 links. CEF
>>loadbalancing will be almost as effective as EtherChannel, while
>>EtherChannel is real difficult to troubleshoot. If you have your mind
>>set on EtherChannel go with L3 one.
>>
>>In most today networks VTP brings no benefit, so I would recommend you
>>to run all you boxes in transparent mode (not in server mode). It is
>>fine then to use same domain name. If you think you will benefit from
>>VTP let me know why you need it, and I may be able to give you another
>>recommendation.
>>
>
>Do you feel VTP is any more useful with Version 3? I will admit I
>originally thought of it as a compatibility tool, when you mixed ISL,
>LANE, 802.10 SDE, etc.
>
>_______________________________________________________________________
>Please help support GroupStudy by purchasing your study materials from:
>http://shop.groupstudy.com
>
>Subscription information may be found at:
>http://www.groupstudy.com/list/CCIELab.html
>
>_______________________________________________________________________
>Please help support GroupStudy by purchasing your study materials from:
>http://shop.groupstudy.com
>
>Subscription information may be found at:
>http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sat Jul 03 2004 - 19:40:32 GMT-3