- Test - lists.nanog.org

Re: C&W Peering
by Sean Donelan 07 Jun '01

07 Jun '01

The loss of one or two (or as someone once calculated up to 40%) of the root servers has no or little effect. DNS is relatively robust, but not indestructible. However, no one outside of Cable & Wireless knows just how many networks or even which networks they eventually will cut-off. If C&W continues down the road of "de-peering", they may eventually cut off too much and cease to be a useful Internet provider. They may be a fine private network, but if you eliminate enough interconnectivity you aren't in the Internet any more. PSI was simply the most public case. But it doesn't appear that even Cable&Wireless's own sales force knows how bad C&W's network connectivity is going to get until after C&W finishes cutting off all the networks. There have been private reports that C&W sent out letters to several other providers. Will C&W still be a viable network after its management finishes? I don't know, I don't know if they know. They didn't seem to understand the effect cutting PSI would have. On Wed, 06 June 2001, Rafi Sadowsky wrote: > > Combine the loss of C.ROOT and the loss to E.ROOT for their loss of > > peering to NASA that happened last week and will happen again in a few > > weeks and they may have many issues related to this decision. > > Maybe I'm missing something obvious but I can't help wondering why given > the redundancy of the DNS system the loss of connectivity to 2 or 3 > root-servers should have a significant effect on an ISP who is otherwise > well connected and probably has a low latency/loss link to at least 2-3 of > the remaining root DNS servers > > Anyone care to enlighten me ?

1 0

Re: New peering criteria
by Sean Donelan 07 Jun '01

07 Jun '01

On Wed, 06 June 2001, Vijay Gill wrote: > On 6 Jun 2001, Sean Donelan wrote: > > > If you think it will hurt the other provider more than you, you'll > > terminate the agreement. If you think it will hurt you more, you'll > > fight to keep it in place. > > If it will hurt you more than it will hurt them, then you aren't a really > a "peer", and in that case, the disparity is resolved by the use of a > service contract whereby you pay them money in exchange for connectivity. I agree, but how do you decide who is hurt more? And therefore who should be the vendor and who is the customer? Is the "bigger" network always the vendor, or is the network with more content the vendor, or the network with more eyeballs the vendor? That's what I don't understand about the "balance" requirement. Ok, so you know the traffic is imbalanced, but whose fault/hurt is it when traffic is imbalanced? And who is responsible for "fixing" the imbalance in traffic? The simple answer is I'm the vendor and you are the customer, so you should pay me. The more difficult answer is one side turns off the connection (C&W) and the first side to blink is the customer (C&W). If C&W didn't feel any pain, why would they turn the peering sessions back on?

6 6

RE: jumbo frames
by Lane Patterson 07 Jun '01

07 Jun '01

Has anybody been running PMTU-D scans across networks from a 4470-connected host? I would love to see some data. I have been tracking jumbo frame support trends for a while, and am reasonably disappointed by lack of standards and vendor willingness to support jumbos (yes there are very REAL h/w design considerations, so until operators demand jumbo support and folks test it in realistic environments, it's not going to happen). Unfortunately, many of the folks most adamant about maintaining 4470 in their core are therefore sticking with POS everywhere, so their requirement is not making it to the ether vendors. There are different reasons to use several different sizing parameters: "Mini-jumbo": say 1518, 1540, etc. the idea here is that you can handle stacked tunnels and LAN encapsulations, such as stacked headers of 802.1Q, MPLS, IP/GRE tunnels, etc. while still preserving "1500 for the edge" Applicability: 802.1Q, VPN, MPLS, and other encap-based or tunnel-based applications "Mid-jumbo": say 4470: the idea is to make sure a backbone can preserve its MTU across both ethernet, ATM, and POS links within its diameter, and conceivably between networks via IX's that support jumbos. This in fact may be critical for folks running large ISIS implementations that need to ration # of LSPs: http://www.ietf.org/internet-drafts/draft-kaplan-isis-ext-eth-02.txt Applicability: -internal: maintain 4470 for router-router control traffic such as ISIS -external: allow for customers that use larger MTU, preserve MTU across IX peering points "Real jumbo": not standardized, but somewhere between 8100-9100B, this is for servers that want to pack GigE links with optimized I/O based on 8K memory chunks, ala the original reasons for Alteon jumbo support. Of course, need for these jumbos is probably still within LAN/MAN scope for the next generation of operational deployment. Applicability: -optimizing server thruput by reducing per-packet overhead, and directly mapping data payload to a memory chunk with no "SAR" buffering function. -scope is LAN/MAN Cheers, -Lane > -----Original Message----- > From: Richard A. Steenbergen [mailto:ras@e-gerbil.net] > Sent: Wednesday, May 30, 2001 3:05 PM > To: Wayne Bouchard > Cc: Dave Siegel; Tony Hain; nanog(a)merit.edu > Subject: Re: jumbo frames > > > > On Wed, 30 May 2001, Wayne Bouchard wrote: > > > Well, the way it oughta work is that the backbone uses the > same MTU as > > that of the largest MTU of your endpoints. So, for example, > you have a > > buncha hosts on a fddi ring running at 4470, you want to make sure > > those frames don't have to get fragmented inside your > network. Idealy, > > all hosts have the same MTU and no one has to worry about > that, but in > > practice, it seems to be better to push the fragmentation > as close to > > the end user as possible. (That is, if a user on a 1500MTU > link makes > > a request to a host on a 4470 link, the response is 4470 up > until the > > user's end network.) Of course, path MTU discovery makes this a moot > > point. The conversation will be held in 1500 byte fragments. > > Fortunantly hosts on FDDI rings are rare these days, but I'd > love to see a > modern analysis of the packet sizes going through the > internet (everything > I've seen comes from the days when FDDI roamed the earth). > > From everything I've seen out of IEEE, they continue to view > Ethernet as a > "LAN Standard" and don't really want to consider its use in > the core, even > for 10GigE. As long as the creation of 99.999% of packets is <= 1500 > bytes, and the links which pass packets are equal or greater, noting > really nasty happens. The argument is that "most people won't really > benefit from it, and it will introduce incompatibilities in > MTU size, so > why should it be a standard", which misses the potential use > in WAN links. > > I don't expect to see many hosts w/10GigE cards for a while, > but it would > be nice if Path MTU Discovery was a bit better. > > -- > Richard A Steenbergen <ras(a)e-gerbil.net> > http://www.e-gerbil.net/ras > PGP Key ID: 0x138EA177 (67 29 D7 BC E8 18 3E DA B2 46 B3 D8 > 14 36 FE B6) >

4 3

Re: Why so little traffic from C&W
by Paul Vixie 07 Jun '01

07 Jun '01

> PSI spent a lot of money over the last few years (perhaps more than C&W > spent), which may also be why PSI is in financial trouble. In the cable > TV industry, cable companies must pay for content (e.g. cable companies > pay CNN for the privilege of transmitting CNN on their network). Since > PSI has more content than C&W (as evidence by the imbalance), why shouldn't > C&W have to compensate PSI for its content just like the cable companies? if psi had to pay its customers to get exclusive access to that content, then sure as hell c&w would have to pay psi (or not, if the content had no end-eyeballs willing to pay c&w for access to it.) i think the reasoning goes something like "your customers pay you to send this stuff out for them, you should share that money with the folks who do your final delivery for you." (i don't agree with the reasoning, but i think i understand the thoughts being thunk.)

1 0

RE: POS OC48 interfaces using different wave length
by pasi.vaananen＠nokia.com 07 Jun '01

07 Jun '01

> >I am thinking about using different POS interfaces on dark-fibers. > > > >On one side, i already have POS-OC48-LR card (Cisco). > >This card use 1550 nm wave length. > > > >We would like to use a POS-OC48-SR (Cisco). > >This card use 1310 nm wave length. > > > >Are 1310 and 1550 optics comptatible ? > > You have to have the same wavelength at both ends, otherwise it won't work. > > >If yes, i guess that there is impact on the power Budget available. How-Much ?! > > LR tends to be 1550nm, because 1550nm has less signal loss over the same > length of fiber. > The receivers are typically wideband, and can receive signals transmitted on either wavelength, 1550nm window is used because of lower signal loss as pointed above. So, provided that the link length is in 1310nm power budget, there should not be harm on using 1550nm other way. In short links, depending on laser types, and overall link attenuation you might need to add attenuator on the link to prevent the saturation in the receiver. In case you have the goods in hand, you can try it out, but use attenuators to emulate link if you are just connecting equipment together over short fiber link. Pasi

1 0

RE: Hosting Failover Question
by Rishi Singh 07 Jun '01

07 Jun '01

5 seconds is very good if you can attain it. I have two Nokia IP650s running with VRRP monitored circuits. The firewalls are constantly trading vrrp packets and firewall table sync state. The fastest failover I've gotten it down to is about 30 seconds with complete stateful failover. This has worked well for my applications, which are financial revenue based trading apps. By the time the trader's screen has frozen, it takes them about 15 seconds to bang on the keyboard, 5 seconds to get up and scream, and 10 seconds to call someone, by the time its already failed over ;-). -----Original Message----- From: James DeMong [mailto:James.DeMong@telus.com] Sent: Wednesday, June 06, 2001 12:20 PM To: Nanog@Merit. Edu (E-mail) Subject: Hosting Failover Question I am looking for a Rule of Thumb on failover for hosting, mostly tcp(webserver), some udp (audio/video streaming) stuff. My thought is that 5 seconds to failover when a network element(switch/router/load balancer/FW) fails is reasonable. How many seconds is reasonable for fail over in such a situation? Thanks in advance. __ James DeMong Network Design Specialist TELUS Advanced Communications Phone: (403) 503-3718 Email: James.DeMong(a)telus.com

1 0

CABLE & WIRELESS STATEMENT
by PETER JANSEN 07 Jun '01

07 Jun '01

CABLE & WIRELESS STATEMENT As part of the review of our current peering arrangements, last Saturday (June 2nd), Cable & Wireless disconnected PSINet's peering connections as they no longer meet our publicly stated peering requirements (www.cw.com/us/peering) Over the last few days, discussions have continued between Cable & Wireless and PSINet to find a resolution to this issue for the benefit of all parties. On June 5 PSINet agreed, through a letter of intent, to meet Cable & Wireless' peering policy. PSINet has come forward with new information that may demonstrate PSINet meets Cable & Wireless' peering policy. As an expression of good faith, Cable & Wireless has reestablished the peering connections with PSINet for a period of 60 days while the parties consider PSINet's new information and, if applicable, negotiate in good faith a new peering agreement. Peter Jansen Cable & Wireless

1 0

Sprint NAP
by Alex Rubenstein 07 Jun '01

07 Jun '01

So, I am trying to find out whats going on with the SprintNAP; I'd heard that they were going to have an ATM platform, yada yada, but whatever. Anyone know anything about it? Who to contact? Pricing?

1 0

Re: And then there were two
by Ron Buchalski 07 Jun '01

07 Jun '01

>From: Sean Donelan <sean(a)donelan.com> >To: nanog(a)merit.edu >Subject: And then there were two >Date: 5 Jun 2001 17:16:13 -0700 > > >If you accept the premise that "peer == equal" does that mean >in the end there will be only two ISPs each with exactly 50% >of the world's Internet because no one else will be an equal? Why can't you have more than two 'equals'? Couldn't you have three 'equals' or four 'equals'? It would be just as difficult to maintain three or four _exact_ divisions as it would be to maintain two. >I've never understood how the word "peer" mutated from its >technical definition arising from its use in the BGP protocol >to its use by marketing people. The keyword is 'marketing' people. Aren't they the ones who always twist and exploit technical terms? >As far as I can tell, EGP (Exterior Gateway Protocol) originally >used the term "neighbor." Berkeley used the berkelism "peer" in >their software and RFC 911 documenting their experience, and the >term stuck through EGP2, BGP1-4. > >If we still used the word "neighbor" would the phrase "Are you >a neighbor?" have a different ring than "Are you a peer?" You >can have lots of neighbors, even if you think you are superior >to all of them. You're thinking about this too much. The backhoes must be behaving today! ;-) -rb _________________________________________________________________ Get your FREE download of MSN Explorer at http://explorer.msn.com

3 3

Self Regulation (was Re: New peering criteria)
by smd＠clock.org 07 Jun '01

07 Jun '01

| I would imagine that such actions tend to irritate the hive of regulatory | officials. Self regulation would be one way to prevent such action. The answer is simple: have more lawyers than most jurisdictions have civil servants. q: what is a phone company? a: it's a law firm with a side-interest in telecommunications q: what do phone company facilities look like? a: boxes with no windows stuffed with lawyers and an antenna on top q: who has more employees? a standalone ISP or a phone company's legal branch's Human Resources department? a: that's not funny. see you in court. q: is it true that some phone companies' legal branches are revenue centres? a: oh yes. and don't forget that at least one IXC _literally_ sued itself into existence. Is there any serious and relevant regulatory authority which is going to intervene in a contractual dispute between parties which are NOT dominant in their industry? It seems to me that players like C&W face vigorous competition. Sean. PS - can you tell how much i hate it when vaguely technical people pose ex cathedra about vaguely legal & regulatory issues? - -- Sean Doran <smd(a)clock.org> >From WordNet (r) 1.6 [wn]: "mulch, noun. 1. a protective covering of half-rotten vegetable matter (cf legal department)"

3 2