Date: Wed, 14 Mar 2001 15:49:03 -0800 (PST) From: "John K. Doyle, Jr." <John.Doyle@oracle.com>
Kevin,
I'm surprised to see YOU (an old DECnet person :)) recommending this. You know he's gonna die if he's running pipes larger than T1s. The process-switching will eat him alive, no?
You can do better than a T-1 with most routers, but I can't argue that it's efficient. If TGV was still around, I'd whole-heartedly recommend it. I should have looked at the current offering from Process. I see that the current version does support DECnet over IP. But even though MultiNet was VERY well written in this area, it is still limited bandwidth. It's just that the bottlenecks are at the end systems and not at the routers. But this is definitely a better way to go. R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: oberman@es.net Phone: +1 510 486-8634
Date: Wed, 14 Mar 2001 15:49:03 -0800 (PST) From: "John K. Doyle, Jr." <John.Doyle@oracle.com>
Kevin,
I'm surprised to see YOU (an old DECnet person :)) recommending this. You know he's gonna die if he's running pipes larger than T1s. The process-switching will eat him alive, no?
You can do better than a T-1 with most routers, but I can't argue that it's efficient. If TGV was still around, I'd whole-heartedly recommend it.
What's the big deal. Put up a one interface box up that has only the job of doing DECnet encapsulation onto the tunnel. It will go as fast as it goes and who cares about the size of the pipes. No reason to mix this with the egress router for the network. It meets the stated design goal of providing transparent DECnet and will not impact anyone else should it run out of cycles. I may draw some heat, but DECnet was built in the days of slow links and limiting an ethernet to a thousand DECnet packets per second (about what I would expect from a small cisco) won't get the users too unhappy. I don't know how fast a fast 7200 would do this, but I'd guess it could do a good deal better than that. How much the users want to pay sets the speed they want to go, sounds right to me. jerry
participants (2)
-
Jerry Scharf -
Kevin Oberman