Chris, I wrestled with this question about a year ago while working with G appliances and here's my take on it. Though I have not played with uplinkfast or backbonefast, I recommended enabling portfast on those ports connected to the G. Here's my reasoning (and it's been a while since I've done this, so anyone feel free to correct me :)
Spannning tree's main purpose is to avoid switching loops within your network. Now, if you take a fully functioning network with IPS working and you don't have switching loops, then you just need to worry about when link-state goes down on the G for that split second. If there are no loops created during that period of time, you can bring the G back up w/o waiting through the STP learning period and have no problems. Since the period of time that the link-state is actually down is fairly small (less than a second IIRC), you're betting that no one will create a switching loop on another part of your network during that time. Since connectivity after a G power loss was a paramount concern, I recommended using portfast to bring those switch ports up as fast as possible. Your environment may have different requirements, so as usual YMMV. -Matt -- Matt Kaar, CISSP Carnegie Mellon University Information Networking Institute [EMAIL PROTECTED] On 6/24/05, Chris Norris/AMIG <[EMAIL PROTECTED]> wrote: > > Just curious to know how others are handling the physical install of > Inline > IDP devices. We are looking to move our Proventia G to inline mode as it > wasn't installed this way originally. > > I was told that in "acts just like a cable" and would not be a problem in > passing traffic in the event of a power failure on the device. That's not > exactly true. With no power it passes traffic "like a cable" but when > power > is present the IDP establishes link-state with the 2 switches it's > connected to. When power is lost so is link-state with the switches which > can invoke a spanning-tree change. > > This is what happened during our test. It's too bad that the device is not > truly "passive" from a link-state perspective so that it would allow the > switches to "see" each other through the IDP, but it is what it is. > > So my suggestion to our network team is to look at options such as > "uplinkfast" or "backbonefast" since they are using Cisco switches. I > suppose they could use "portfast" on the IDP ports but I have always > frowned on "portfast" (which disables Spanning-Tree learning mode) on > anything but end user ports. > > What are other people doing? > > Regards, > Chris Norris CISSP > American Modern Insurance Companies > Sr. Security Engineer > IS Risk and Security Management > > _______________________________________________ > ISSForum mailing list > [email protected] > > TO UNSUBSCRIBE OR CHANGE YOUR SUBSCRIPTION, go to > https://atla-mm1.iss.net/mailman/listinfo/issforum > > To contact the ISSForum Moderator, send email to [EMAIL PROTECTED] > > The ISSForum mailing list is hosted and managed by Internet Security > Systems, 6303 Barfield Road, Atlanta, Georgia, USA 30328. > _______________________________________________ ISSForum mailing list [email protected] TO UNSUBSCRIBE OR CHANGE YOUR SUBSCRIPTION, go to https://atla-mm1.iss.net/mailman/listinfo/issforum To contact the ISSForum Moderator, send email to [EMAIL PROTECTED] The ISSForum mailing list is hosted and managed by Internet Security Systems, 6303 Barfield Road, Atlanta, Georgia, USA 30328.
