the iab simplifies internet architecture!
reported from tonight's iitf iab (internet archetecture board) plenary. proclaimed by an esteemed iab member from the podium: "it is bad in the long term to add hierarchy to routing" this will save a lot of work. whew! randy
On Thu, Nov 10, 2005 at 07:10:13PM -1000, Randy Bush wrote:
reported from tonight's iitf iab (internet archetecture board) plenary. proclaimed by an esteemed iab member from the podium:
"it is bad in the long term to add hierarchy to routing"
this will save a lot of work. whew!
randy
url for the stream? i -have- to see this ... --bill
"it is bad in the long term to add hierarchy to routing" url for the stream? i -have- to see this ...
reported verbatim separately by two friends who have routing clue but not enough clue to stay away from the iitf. so you may just have to wait. but it will be a classic. if you can get and edit it, send it to boing boing or /. randy
Maybe Bob Braden's presentaion in e2e task group could do some help. In fact, they just start to discusss what will be the next generation architecture, but does not reach agreement at all. http://www.isi.edu/~braden/e2e-tf/braden.newarch.ppt Joe --- Randy Bush <randy@psg.com> wrote:
"it is bad in the long term to add hierarchy to routing" url for the stream? i -have- to see this ...
reported verbatim separately by two friends who have routing clue but not enough clue to stay away from the iitf. so you may just have to wait.
but it will be a classic. if you can get and edit it, send it to boing boing or /.
randy
__________________________________ Do you Yahoo!? New and Improved Yahoo! Mail - 1GB free storage! http://sg.whatsnew.mail.yahoo.com
but it will be a classic. if you can get and edit it, send it to boing boing or /. Pearls before swine.
that's what a number of i* members have publicly stated is their opinion of talking to us operators. i saved in my mementos the following quote from an ipv6 architect and current iab member, "operators won't accept the h ratio because they don't know what a logarithm is."
In my rss aggregator, boingboing and /. are labeled
i.e. you read them personally i find doctorow on boing boing often quite fun. /. has high n:s but not all that much higher than this forum. and i suspect i would count this message as n. apologies. randy
None that I have spoken with. What I hear continually is that people would like operational viewpoints on what they're doing and are concerned at the fact that operators don't involve themselves in IETF discussions. On Nov 11, 2005, at 6:03 AM, Randy Bush wrote:
that's what a number of i* members have publicly stated is their opinion of talking to us operators.
On Fri, Nov 11, 2005 at 07:39:09AM -0800, Fred Baker wrote:
None that I have spoken with. What I hear continually is that people would like operational viewpoints on what they're doing and are concerned at the fact that operators don't involve themselves in IETF discussions.
Agreed, but it is pretty clear that serious communication/image/respect/etc challenges remain. That is why the current IAB took a step, albeit a small step, towards trying to change that by opening up the communication channels a bit. As I seem to become fond of saying, "its a first step, but you have to start somewhere". I'm hoping (and pushing) that we keep moving in this direction. As always, we can all educate each other a bit, and some of that happened at the NANOG BOF. However, much more is needed. To that end, I've applied for a slot at APRICOT for the IAB so we can keep what momentum we gained from the NANOG BOF going. Finally, if folks have suggestions as to how to make these (communication) channels more useful, work better, etc (including suggestions for issues you'd like to talk to the IAB about or hear the IAB talk about), please let me know. Dave n Nov 11, 2005, at 6:03 AM, Randy Bush wrote:
that's what a number of i* members have publicly stated is their opinion of talking to us operators.
None that I have spoken with.
that's what a number of i* members have publicly stated is their opinion of talking to us operators.
i imagine you speak with the one i was quoting rather often, though you were not there when it was said. i was. ask others who were there, pitsburgh ietf, a meeting between ipv6 chairs, iesg members, rirs, and a few ops. a current member of the iab specifically said, and i quote again, since you seem to have missed the rest of my paragraph, operators won't accept the h ratio because they don't know what a logarithm is. while the ietf mouths a lot of words about wanting to hear from, and get participation from, operators, the actual experience is pretty brutal. http://rip.psg.com/~randy/051000.ccr-ivtf.html is from the current issue of acm sigcomm's ccr, where aaron falk also has a piece. i play curmudgeon and he pollyanna. randy
yes, a specific member of the IAB said that. A few moments ago, I was chatting with the chair of the IAB, who wondered out loud whether he had noticed everyone else on the IAB edging away from him (something about lightning strikes emanating from the dagger-eyes of fellow IAB members I think) and observing that in the viewpoint he was on his own. But your comment was not "PN, member of the IAB, said something clueless that the rest of the IAB disagreed with", nor did your subsequent comment On Nov 11, 2005, at 6:03 AM, Randy Bush wrote:
but it will be a classic. if you can get and edit it, send it to boing boing or /.
Pearls before swine.
that's what a number of i* members have publicly stated is their opinion of talking to us operators.
distinguish between the IAB, the IESG, the IETF, ISOC, and or any of the other acronyms that start with the letter I. Yes, the experience of communicating between groups with different expertise can be brutal, and the brutality goes both directions. A classic example relates to discussions I have with various military agencies and developing countries on their issues, and when I pursue solutions to same get comments from some members of this community (note the lack of broad-brush over-generalization) that "we don't need that so it is a stupid idea". Well, if one is running a static fiber core and has effectively infinite bandwidth everywhere with very high reliability, it probably is. It's hard to run fiber to a geosynchronous satellite - that's a lot of glass, at a minimum. I would suggest that we drop the overgeneralizations, in which "PN" becomes "The I*", drop the disrespectful associations ("pearls before swine"), and drop the tone. Guys, we're all in this together, and it would be better if we spent a nanosecond thinking about how to get along. On Nov 11, 2005, at 10:42 AM, Randy Bush wrote:
None that I have spoken with.
that's what a number of i* members have publicly stated is their opinion of talking to us operators.
i imagine you speak with the one i was quoting rather often, though you were not there when it was said. i was. ask others who were there, pitsburgh ietf, a meeting between ipv6 chairs, iesg members, rirs, and a few ops. a current member of the iab specifically said, and i quote again, since you seem to have missed the rest of my paragraph,
operators won't accept the h ratio because they don't know what a logarithm is.
while the ietf mouths a lot of words about wanting to hear from, and get participation from, operators, the actual experience is pretty brutal.
http://rip.psg.com/~randy/051000.ccr-ivtf.html is from the current issue of acm sigcomm's ccr, where aaron falk also has a piece. i play curmudgeon and he pollyanna.
randy
[ many folk may wish to skip to the *** ]
yes, a specific member of the IAB said that.
and we have let their name live in peace. and my message made it very clear that it was one member speaking.
wondered out loud whether he had noticed everyone else on the IAB edging away from him (something about lightning strikes emanating from the dagger-eyes of fellow IAB members I think) and observing that in the viewpoint he was on his own.
actually, i was hoping that there was a coherent vision behind the provocative statement that the speaker would be willing to share with nanog and work on developing. the ops community is absolutely *desperate* for a real vision of how we move forward for the long term in addressing and routing with realistic technologies and actually viable transition strategies. and we sure have not seen them yet.
On Nov 11, 2005, at 6:03 AM, Randy Bush wrote:
but it will be a classic. if you can get and edit it, send it to boing boing or /. Pearls before swine. that's what a number of i* members have publicly stated is their opinion of talking to us operators. distinguish between the IAB, the IESG, the IETF, ISOC, and or any of the other acronyms that start with the letter I.
fred, you're talking to someone who was stupid enough to waste a dozen years of his life tilting at those very windmills. i have been attacked as a clueless operator, and heard operators as a class denigrated, by each and every one of them [0]. apologies that i did not have the bad taste to keep other than the most amusing quotes.
Guys, we're all in this together, and it would be better if we spent a nanosecond thinking about how to get along.
we have spent decades trying to get along, and will continue to do so. we don't have a choice, as we're all in the same bathtub. but in my experience, there are very different goal sets, means of achieving them, etc. and whitewashing over the problems with "let's play nice," "be a team player," and "charlene is not playing nice and there is no real problem," just makes the problems fester longer and deeper and makes them worse when they inevitably force their way to the surface and explode. denial may not be just a river in egypt, but the erosion is serious in all its instantiations. *** but credit to leslie, who did come to nanog (and has been at many ops fora) and did listen. though the discussion was a bit frank, which, of course, it never is at ivtf <sound of pigs wings>, i think the message was clear, though far from new. and she has my sympathies for what i imagine (having tried to communicate the same message for a while) she faced when she tried to bring it back to the ivtf. likely she will has more skill at this than i, no big surprise there. and nanog really appreciated dave's ten minute description of who the ivtf and iab are. we ops natives are soooo uneducated. i suggest, as opposed to the forever chant "operators should come to ivtf and participate," that more ivtf folk should come to operational fora and try to participate. with the ivtf's move to mediocrity and complexity, it is no longer the center of the universe. and come as equals, not as the enlightened slumming. we're all just bozos on this bus. we need to meet where the rubber meets the road, not the sky. if there is something nanog can do to facilitate this, shout. and i am sure the same goes for other ops fora. but please don't plan yet another "the wonderful things the ivtf is doing in area x." try something more like "what are the most critical forward problems and what are the deployment, transition, and use constraints on possible approaches?" the goal is / would be to develop technology that is actually useful and deployable and with realistic transition plans. and no it will not be easy. but denial does not make it easier. randy --- [0] - just yesterday, i wore my "bottom feeding scum sucker" tee shirt from ivtf, i think, summer '95.
have been attacked as a clueless operator, and heard operators as a class denigrated, by each and every one of them [0].
randy
---
[0] - just yesterday, i wore my "bottom feeding scum sucker" tee shirt from ivtf, i think, summer '95.
mar1996 - Los Angeles. CIDRd working group. only a dozen such shirts ever were printed. (altho i still have the artwork). operators can poke fun at themselves... the only non-op to get a shirt has passed on. --bill
CIDRd working group.
ahh yes. a memorable period of openness, cooperation, and respect for operators in the ivtf community.
i still have the artwork
i always loved the baby diaper yellow shirt color far more than the barely decipherable koi on the back. great color! <http://rip.psg.com/~randy/051111.bfss-tee> that and a deep orange nanog tee are two of my fave plain ones. but the nanog eugene dead/keasey set has to be the best of all. ymmv. randy
but please don't plan yet another "the wonderful things the ivtf is doing in area x."
Actually, that is not at all what I had intened or planned, and if it came across that way then to some extent we failed. In any event, I do appreciate this feedback.
try something more like "what are the most critical forward problems and what are the deployment, transition, and use constraints on possible approaches?"
Excellent suggestions. Much appreciated. Thanks, Dave
On Fri, 11 Nov 2005, Randy Bush wrote:
i suggest, as opposed to the forever chant "operators should come to ivtf and participate," that more ivtf folk should come to operational fora and try to participate. with the ivtf's move to mediocrity and complexity, it is no longer the center of the universe. and come as equals, not as the enlightened slumming. we're all just bozos on this bus.
What exactly is there [for vendors] to participate in at operational fora? That is, except for getting general clue on what's going on -- and putting a pinch of salt as the view one sees may not necessarily be representative. It's a two way street; vendors need to listen to the ops folks. Ops folks need to participate in the IETF. Yeah, it's often quite frustrating, especially when the vendors have already "packed the rooms" and most ops who came around {don't want to, cant'} spend energy fighting that. -- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
That is, except for getting general clue on what's going on -- and putting a
But, um, isn't that pretty important? Many of the NANOG talks are specifically geared to let vendors know what operators need.
pinch of salt as the view one sees may not necessarily be representative.
How many of ours are?
It's a two way street; vendors need to listen to the ops folks. Ops folks need to participate in the IETF. Yeah, it's often quite frustrating, especially when the vendors have already "packed the rooms" and most ops who came around {don't want to, cant'} spend energy fighting that.
Pekka, give NANOG (the meetings) a try sometime. You might like it, and we're very eager for your input.
On Mon, 14 Nov 2005, Randy Bush wrote:
It's a two way street; vendors need to listen to the ops folks.
because they want to sell their equipment and software to the operators?
yes, including improving (in various ways) their existing equipment and software to make the customer happier.
Ops folks need to participate in the IETF.
because they want to sell what? clue? seems unmarketable.
So that they can affect the protocols that are going to be implemented at a stage where they can still be modified to suit their needs, scenarios, requirements, etc. Options for changing the protocols are somewhat more limited (though not zero) when the specs and code (those that don't address the needs of a particular set of operators as-is, in any case) have already shipped. -- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
Ops folks need to participate in the IETF. because they want to sell what? clue? seems unmarketable.
So that they can affect the protocols that are going to be implemented at a stage where they can still be modified to suit their needs, scenarios, requirements, etc.
Options for changing the protocols are somewhat more limited (though not zero) when the specs and code (those that don't address the needs of a particular set of operators as-is, in any case) have already shipped.
i think ops folks have leverage in slightly different ways. for commondity products, i expect you are right. for high end products, i have found that when i ask a vendor for certain features, and am willing to give them money, they tend to treat those feaure requests with favor. features may or may not be reflected as IETF or other SDO based specifications. in fact, recent events lead me to believe that marketing groups from vendors, having done some market/customer research, are presuring the engineeering groups to push certain specs in the IETF. To have the actual customers show up at the IETF and attempt to influence the specs directly will put the engineering folks in a bind. ... believe the customer or the marketing department? if a vendor ships code that does not meet my needs or is harmful to my operations, i will either dump the vendor or encourage them to build to my needs, regardless of the IETF. --bill
-- Pekka Savola "You each name yourselves king, yet the Netcore Oy kingdom bleeds." Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings
It's a two way street; vendors need to listen to the ops folks. because they want to sell their equipment and software to the operators? yes, including improving (in various ways) their existing equipment and software to make the customer happier.
somehow, the vendors hear from their customers. the problem is that isp operators are a small portion of the market. so having enterprise, r&e, gummint, ... at nanog is good as it impacts a larger target area in the vendors. when we all agree and mount max pressure, we maybe have a 50% success ratio.
Ops folks need to participate in the IETF. because they want to sell what? clue? seems unmarketable. So that they can affect the protocols that are going to be implemented at a stage where they can still be modified to suit their needs, scenarios, requirements, etc.
and what success ratio have we had with this? 10%? look at the poster child for operator success at the ietf, ipv6. it took seven years to get rid of tla/nla etc. it took eight+ to get rid of site-local, yet it keeps rearing its head. and we still don't have squat for routing. randy
Randy Bush wrote:
but it will be a classic. if you can get and edit it, send it to boing boing or /. Pearls before swine.
that's what a number of i* members have publicly stated is their opinion of talking to us operators. i saved in my mementos the following quote from an ipv6 architect and current iab member, "operators won't accept the h ratio because they don't know what a logarithm is."
I'd like to see that. Should prove good fodder for quotes on slides. /vijay <-- winner of several math prizes
btw, for another great giggle (many thanks to brian candler for reporting it) From the documentation for Cisco's VPN client software for Linux: http://www.cisco.com/en/US/products/sw/secursw/ps2308/products_user_guide_ch... "User profiles [which contain all your IPSEC parameters: pre-shared key, username and password] reside in the /etc/CiscoSystemsVPNClient/Profiles/ directory. Leave the permissions for the Profiles folder set at drwxrwxrwx. Each profile in the Profiles folder should have the follwoing permissions: -rw-rw-rw-." i think we may have hit an elbow in the entropy curve. heat death of the internet predicted. news at 11:00 gmt. randy
On Nov 10, 2005, at 9:14 PM, bmanning@vacation.karoshi.com wrote:
On Thu, Nov 10, 2005 at 07:10:13PM -1000, Randy Bush wrote:
reported from tonight's iitf iab (internet archetecture board) plenary. proclaimed by an esteemed iab member from the podium:
"it is bad in the long term to add hierarchy to routing"
this will save a lot of work. whew!
randy
url for the stream? i -have- to see this ...
--bill
Bill, I was there at the IAB plenary. I am afraid that the above quote was inaccurate and out of the context. Lixia
At 7:10 PM -1000 11/10/05, Randy Bush wrote:
reported from tonight's iitf iab (internet archetecture board) plenary. proclaimed by an esteemed iab member from the podium:
"it is bad in the long term to add hierarchy to routing"
this will save a lot of work. whew!
That is exceptionally good news! It follows that all IPv6 assignments for end-users can therefore be a /64... ;-) /John
On Fri, Nov 11, 2005 at 04:29:33PM -0500, John Curran wrote:
At 7:10 PM -1000 11/10/05, Randy Bush wrote:
reported from tonight's iitf iab (internet archetecture board) plenary. proclaimed by an esteemed iab member from the podium:
"it is bad in the long term to add hierarchy to routing"
this will save a lot of work. whew!
That is exceptionally good news!
It follows that all IPv6 assignments for end-users can therefore be a /64... ;-) /John
"bridge where you can, route where you must." -- i forgot where this came from? Radia? ARP. ARP. arp,arp,arp,arp.... <bcast storm>. --bill
I believe that it is attributable to John Hart, Vitalink, late 1980's. If he didn't coin it, he sure quoted it a lot. Radia would have said something more like "bridge within a campus and route between them", I suspect. On Nov 11, 2005, at 1:36 PM, bmanning@vacation.karoshi.com wrote:
"bridge where you can, route where you must." -- i forgot where this came from? Radia?
On Mon, Nov 14, 2005 at 05:53:04AM -0800, Fred Baker wrote:
I believe that it is attributable to John Hart, Vitalink, late 1980's. If he didn't coin it, he sure quoted it a lot.
Radia would have said something more like "bridge within a campus and route between them", I suspect.
On Nov 11, 2005, at 1:36 PM, bmanning@vacation.karoshi.com wrote:
"bridge where you can, route where you must." -- i forgot where this came from? Radia?
there was 3com inthere somewhere too. (fond memories of Vitalink and a globally bridged network. ARP storms have a special place in my heart) the counter phrase was, of course, "route where you can, bridge where you must". --bill
participants (13)
-
bmanning@vacation.karoshi.com
-
David Meyer
-
Fred Baker
-
Joe Shen
-
John Curran
-
Lixia Zhang
-
Pekka Savola
-
Randy Bush
-
Robert E.Seastrom
-
sob@harvard.edu
-
Susan Harris
-
Tony Li
-
vijay gill