June 1998 - Test - lists.nanog.org

Re: different thinking on exchanging traffic
by Richard Mataka 04 Jun '98

04 Jun '98

As an interested observer who created NY IIX in New York I can tell you that getting a NAP or peering point off the ground is no easy task. One of the biggest obstacles that I have seen (IMHO) is education. There are just soooo many ISPs around who do not understand what peering is or what advantage peering can be for them!! The concept of "if you build it will they come" works in some cases but not all. With NY IIX the original idea was for it to be an international exchange point but then even regional backbone providers wanted to peer. Now we have all mixes from international to regional and many private peers. However even with close to 40 internet companies all at one location, only about 10 are currently taking advantage of the NY IIX NAP. Many of the rest are doing private peering. I know this will increase but its a wait and see attitude. I think that as the net is maturing, there are definite needs for the local NAPs, regional NAPs and then National NAPs. Working on a tiered approach may be what is needed to alleviate overloads at the MAEs and evenly distribute traffic. However, as I stated above, for this to work, education is the key as the local ISPs need to keep traffic local and etc. etc. etc... I hope that everyone doesn't mind my 1 cents worth on this thread!!!!! Regards Richard Mataka

1 0

None
by Sean Butler 04 Jun '98

04 Jun '98

John Fraizer wrote: >The thing that makes it "interesting" is the fact that most implementations >DO send an ICMP unreach back. The ICMP Unreach traffic alone generated in >the neighborhood of 1.7Mb before they routed the netblock in question to a >loopback interface on the 7507. The attacker was sending less that 300Kb >of traffic and consuming 2Mb. Any idea where that much amplification is coming from? For smurf with an echo request to a broadcast, its easy to see why there is so much amplification. But for a TCP or UDP packet to port 0, wouldn't just one port unreachable be sent back to the (spoofed) source? Or is it a broadcast TCP or UDP packet to port 0 ??? Thanks, Sean Butler, IBM Global Services

1 0

US West and RADSL
by Marianne Granoff 04 Jun '98

04 Jun '98

>On Wed, 3 Jun 1998, Jonathan Arneault wrote: > >> I just read (http://www.internews.com/isp-news/1998/06/0301-bell.html) that >> Bell Atlantic is looking to finally roll out xDSL. However, not as a >> wholesale, unbundled service, but as an Internet service which will directly >> compete against ISPs. >[snip] >> for us "little guys" to compete against the behemoths? Has anyone had >> similar experiences with their primary iLEC (such as PACBell and US West)? > >US West just rolled out RADSL in Utah. > >$40/month for a 256k bi-directional DSL line > >USWEST.NET, US West's ISP, will sell you dedicated internet access over >that connection for $19.95/month. > >Of course, the number to order either service is the same. > >Of course when you call to order the DSL line they might give you a hard >sell to also sign-up with USWEST.NET. > >Of course they might give out inconsistent and false information that >changes with every time you talk to them. > >We are a DSL enabled ISP. Our first DSL customer ordered his DSL line and >told them he wanted us as his ISP. When the installed his line, they >mapped him to USWEST.NET's ISP!!! I also know of two other confirmed >cases were the customer chose a local DSL ISP, and they mapped them >instead to USWEST.NET. > >I could go on and on about the VERY APPALLING situation here with USW and >DSL and Internet access, about regulated and unregulated services, etc, >etc. > >Dax Kelson >Internet Connect, Inc. The preceeding message was forwarded to me by one of our NM ISPs. US West has been co-marketing its Internet service _with_ its RADSL service in all of its 14 states (http://www.uswest.com/com/customers/interprise/dsl/) Actions by ISP groups in Oregon and New Mexico may provide some relief to local ISPs there. I have just put up a listserve for ISPs in the US West states to use in sharing information. Technet has had one for the NM ISPs for several years - it has been a big help in getting out the word about some of these actions. How about uswisp(a)lists.nm.net? Please feel free to send this out to any interested ISPs. To subscribe, just send an empty note to: uswisp-subscribe(a)lists.nm.org I believe that all the RBOCs/ILECs have taken or will be taking similar actions. Even Sprint's new ION services are part of this trend. In my opinion, this is anti-competitive behavior by monopoly organizations. I think that many local ISPs will be severly hurt by such actions, and more than a few will close their doors. As I see it - the biggest problem is that local ISPs are not organized and do not know how - or have the forums - to work together to fight actions by a company the size of US West (or other RBOC/ILEC). My company, New Mexico Technet, is one of the larger ISPs in NM. We wholesale Internet access to other ISPs. We have intervened in the NM tariff filing for US West's Megabit services (see http://www.technet.nm.org/press.htm) to attempt to correct some of the things that are very anti-competitive about the proposed tariff. So far it has cost us over $30,000 in legal fees and we have not even had the hearing yet. Most local ISPs cannot do this. Most local ISPs do not know how to take the actions with the FCC or with the state public regulatory agencies so that their concerns can even be heard. Frankly - most ISPs are not members of CIX or of ISP/C - and many of them do not even know about those organizations, or understand why they should care. In NM, the local ISPs come in mostly 2 flavors: those that serve urban areas (Albuquerque, Santa Fe, Las Cruces, Los Alamos) who "may" be affected by what US West does but are not sure, and those in rural areas who do not feel they will ever be affected by these actions, and so do not care. In much of US West's 14 state territory - I suspect that this is similar. The local ISPs in the urban areas have mostly seen the other local ISPs as competitors, not collaborators. They have seen US West (or the RBOC/ILEC) as a vendor, not as a competitor. Most local ISPs worry about retribution from US West (or other RBOC/ILEC) (delayed service, unresponsiveness on outages, unfilled orders) if they come on too strong in criticizing a company that they are _so_ dependent on. I am not sure that their concerns are not valid. The local (state) regulatory agencies are overworked and underfunded in this age of telecommunications transition/revolution. It is not that they don't care. It is that they simply have too much on their plates already. Few states provided extra funding to handle all the _new_ issues raised by the 1996 Telecommunications Act at the _state_ regulatory agency level. No one is championing any of this in most states. I think probably because it is not considered a problem by the vast majority of ISP _customers_. I had one of my customers tell me to "just get out of the way" and let US West introduce the high speed service because the customer needed it right away and I was just holding it up. They never saw that US West owned some of the blame in the constant delays, counterfilings, interrogatories, motions to compel, and other actions that have caused this intervention to drag on. Unfortunately, this person is more typical of ISP customers than local ISPs want to believe. The saddest aspect of this is that unless something changes, US West and the other RBOCs/ILECs will likely dominate the supply of Internet access in large urban areas in a few years - and the rural areas will have a great deal less Internet access than they do today. I think the answer is that the local ISPs _and their customers_ have to come together if they want to have choices about ISPs in the future. It will take some of the larger ISPs reaching out to the smaller local ISPs to help them get _all_ of their respective customers informed of the issues. It will take the larger ISPs intervening in more state and FCC proceedings. It will take constant email, listserves, and newsgroups spreading such information - and reaching customers - not just ISPs. It will take some national politicians to "champion" this cause - some who are not worried about losing RBOC/ILEC campaign contributions - which are considerable. It will take involvement by media organizations that are not worried about losing the RBOC advertising revenues - which are also considerable. It will take every local ISP who is harmed by US West actions calling, writing, or emailing their local and national politicians and letting them know that they have informed all of their own customers about the actions by US West or other RBOC/ILEC and informing their customers of the fact that the politician has not responded to these illegal/unethical/anti-competitive actions. In NM, it is now other internet professionals and businesses that have joined ISPs in questioning the actions of US West. Web designers, web page hosting services, internet trainers, web-advertising services and other businesses are starting to realize that US West wants to take their Internet-based business as well. This is a start. Regards, Marianne Marianne Granoff Director of Operations New Mexico Technet, Inc. 5921 Jefferson NE Albuquerque, N.M. 87109 Ph: (505) 345-6555 FAX: (505) 345-6559 email: granoff(a)nm.net or granoff(a)technet.nm.org

1 0

Re: different thinking on exchanging traffic
by shields＠crosslink.net 04 Jun '98

04 Jun '98

In article <Pine.LNX.3.95.980604003057.723W-100000(a)tarkin.fdt.net>, Jon Lewis <jlewis(a)inorganic5.fdt.net> wrote: > I went one step further and talked to some of the local ISPs about pooling > some $ and having a NAP-owned nnrp server. I don't understand the value of having it be NAP-owned vs. being owned by a third party and located at the NAP. -- Shields, CrossLink.

1 0

ADSL & iLEC
by Jonathan Arneault 04 Jun '98

04 Jun '98

I just read (http://www.internews.com/isp-news/1998/06/0301-bell.html) that Bell Atlantic is looking to finally roll out xDSL. However, not as a wholesale, unbundled service, but as an Internet service which will directly compete against ISPs. While I do believe that BA and the rest of the iLECs have the right to compete in their market, I don't think they should have the right to put private, unsubsidized businesses out of business by using their utility based advantages. Additionally, how is it that a "regulated monopoly" is able to transit service outside of the LATA (such as is required with Internet service)? IS this Bell's way of getting into LD, through packet switching? Does anyone know, now that BA has announced that they will provide ADSL internet services, if they provide a similar, tarriffed, unbundled xDSL service by which a local, regional or national ISP could provide a competitively priced service? How does a ten million dollar a year company serve in a market with a 10 billion dollar a year company? Is there a way for us "little guys" to compete against the behemoths? Has anyone had similar experiences with their primary iLEC (such as PACBell and US West)? Hoping for good news, Jonathan Arneault Director of Business Development CMA / INet Solutions Division 518.783.9003 x 253

4 3

Re: Attack/DoS
by John Fraizer 04 Jun '98

04 Jun '98

At 11:38 PM 6/3/98 -0400, Perry E. Metzger wrote: > >"Todd R. Stroup" writes: >> Don't know if it is just me. But over the last 10 hours we have been >> seeing attacks on port 0 from port 0 (both tcp and udp) on several clients >> networks. I have also seen the same attack on port udp 53(DNS). >> >> Anyone have any information on this? > >What do you mean by an "attack"? Are you being flooded? Are the >packets somehow "interesting"? Without details the information is >useless. > >Port 0, btw, is not generally valid, and most proper TCP and UDP >implementations will just send an ICMP Unreachable back when they get >such a packet. > >Perry Perry, Here are some logs of slightly different format that show the same attack Todd writes about: Jun 3 17:02:47 eth0-core0 kernel: IP acct in eth0 UDP 199.199.125.28:53 209.115.17.67:53 L=57 S=0x00 I=47916 F=0x0000 T=49 Jun 3 17:02:47 eth0-core0 kernel: IP acct out eth2 UDP 199.199.125.28:53 209.115.17.67:53 L=57 S=0x00 I=47916 F=0x0000 T=48 Jun 3 17:02:47 eth0-core0 kernel: IP acct out eth0 ICMP/3 209.115.17.65 199.199.125.28 L=119 S=0xC0 I=63767 F=0x0000 T=64 Jun 3 17:02:47 eth0-core0 kernel: IP acct in eth0 UDP 165.113.1.73:53 209.115.17.66:53 L=56 S=0x00 I=25895 F=0x0000 T=57 Jun 3 17:02:47 eth0-core0 kernel: IP acct out eth2 UDP 165.113.1.73:53 209.115.17.66:53 L=56 S=0x00 I=25895 F=0x0000 T=56 Jun 3 17:02:47 eth0-core0 kernel: IP acct out eth0 ICMP/3 209.115.17.65 165.113.1.73 L=118 S=0xC0 I=63769 F=0x0000 T=64 Jun 3 17:02:48 eth0-core0 kernel: IP acct in eth0 UDP 166.93.1.3:63098 209.115.17.66:53 L=56 S=0x00 I=44767 F=0x0040 T=245 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth2 UDP 166.93.1.3:63098 209.115.17.66:53 L=56 S=0x00 I=44767 F=0x0040 T=244 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth0 ICMP/3 209.115.17.65 166.93.1.3 L=118 S=0xC0 I=63770 F=0x0000 T=64 Jun 3 17:02:48 eth0-core0 kernel: IP acct in eth0 UDP 198.81.19.238:4569 209.115.17.66:53 L=59 S=0x00 I=34977 F=0x0000 T=20 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth2 UDP 198.81.19.238:4569 209.115.17.66:53 L=59 S=0x00 I=34977 F=0x0000 T=19 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth0 ICMP/3 209.115.17.65 198.81.19.238 L=121 S=0xC0 I=63771 F=0x0000 T=64 Jun 3 17:02:48 eth0-core0 kernel: IP acct in eth0 UDP 128.112.129.15:56224 209.115.17.66:53 L=58 S=0x00 I=50842 F=0x0040 T=247 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth2 UDP 128.112.129.15:56224 209.115.17.66:53 L=58 S=0x00 I=50842 F=0x0040 T=246 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth0 ICMP/3 209.115.17.65 128.112.129.15 L=120 S=0xC0 I=63772 F=0x0000 T=64 Jun 3 17:02:48 eth0-core0 kernel: IP acct in eth0 UDP 158.152.1.81:53 209.115.17.67:53 L=56 S=0x00 I=21310 F=0x0000 T=53 Jun 3 17:02:48 eth0-core0 kernel: IP acct out eth2 UDP 158.152.1.81:53 209.115.17.67:53 L=56 S=0x00 I=21310 F=0x0000 T=52 The thing that makes it "interesting" is the fact that most implementations DO send an ICMP unreach back. The ICMP Unreach traffic alone generated in the neighborhood of 1.7Mb before they routed the netblock in question to a loopback interface on the 7507. The attacker was sending less that 300Kb of traffic and consuming 2Mb. ------- John Fraizer (root) | __ _ | The System Administrator | / / (_)__ __ ____ __ | The choice mailto:root@EnterZone.Net | / /__/ / _ \/ // /\ \/ / | of a GNU http://www.EnterZone.Net/ | /____/_/_//_/\_,_/ /_/\_\ | Generation A 486 is a terrible thing to waste...

1 0

RE: Sprint's New ION products
by Pete Farmer 04 Jun '98

04 Jun '98

Nathan Stratton [mailto:nathan@robotics.net] wrote: >Right now most of the move to voice over packet/cell networks is >because people want to avoid termination and USF charges. The pace will >accelerate even more now that port density and port cost is approaching >that of TDM. For Sprint simply to use ATM in their core does not let them get around termination and USF. Only if they use an access mechanism other than standard PSTN do they get around these charges. Unfortunately, it's rather difficult to discern what the access mechanism will be for their 'ION' psudo-offering. Pete ______________________________________________________________________ Peter J. Farmer E-mail: pfarmer(a)strategies-u.com Strategies Unlimited Voice: +1 650 941 3438 201 San Antonio Circle, Suite 205 Fax: +1 650 941 5120 Mountain View, CA 94040 WWW: http://www.strategies-u.com

3 2

Re: PPP over Ethernet?
by woody＠zocalo.net 04 Jun '98

04 Jun '98

The standard for PPP over Ethernet is L2TP. The pre-standard proprietary implementations were L2F and PPTP, from Cisco and Microsoft, respectively. I believe you'll find Internet Drafts on all three. The idea is neither "highly bogus" nor used in association with 10Base-T cross-over cables. It's used to tunnel ATCP, IPCP, IPXCP, et cetera over IP. -Bill Woodcock

2 1

111 8th Ave - NYC
by Jonathan Arneault 04 Jun '98

04 Jun '98

Is anyone else co-located in the MFS node in this building? If so, would you be interested in peering? I've got a big honking' router and an ATM switch with only one private peer into them, and a FDDI - MAE connection that isn't currently talking to anyone. If so, e-me off line. Thanks, Jonathan Arneault Director of Business Development CMA / INet Solutions Division 518.783.9003 x 253

1 0

RE: flat vs non-flat charging
by Mike Trest (In Japan) 04 Jun '98

04 Jun '98

Bill, Jerry, Steve, et. al. I don't think I can over stress that the operational issue is not what the technology can support but what patterns the commercial fiber providers of the future will adopt. IMHO, operating separate channels of a WDM drop-and-insert will is no different from operating separate SONET/SDH drop-and-insert. Furthermore, the operational comparisions of multiple point-to-point links with a staticstically multiplexed link has been explored extensively on this list. However, can one *reasonably* expect that the owners of fiber in the ground to allow dark fiber access for the average (or even larger-than-average) network operator. Will the owners of the fiber allow the network operator to manipulate and re-structure his own drop-and-insert revisions over their dark fiber network? IMHO, the best you can reasonably expect is a dedicated port at both ends of a point-to-point link. If the price is right, you can expect to buy as many of these point-to-point links as you can afford. But physical control and network re-structuring will remain in the hands of the fiber plant owner-operator. Perhaps a network operator can plant dark fiber in a campus or regional deployment for extended WDM use. But I suspect the budget would need to be a lot larger (or you would need a really sweet political deal) to gain national access to dark fiber to implement and control the OADM or WDM equipment. The answers to Jerry's original question and the probability of a network operator truly controlling his own WDM network will come from future commercial practice of the fiber owners. ..mike.. At 09:31 AM 6/1/98 -0400, Bill St. Arnaud wrote: >Steven: > >New municipal and campus WDM gear from Cambrian, Ciena and Nortel makes it >a lot easier and cheaper to deliver data services over separate WDM channels >than to MUX them into one SONET or ATM service. More importantly becuase >the service is data transparent, each customer can manage their own WDM >channel independent of the other channels. Try doing that with SONET or >ATM!! > >Check out www.cambrian.com for more details. > >However, I agree with you that if you have spare fiber it is cheaper just to >light up the fiber without WDM or SONET. Packet Engines claims they can >drive 22km of dark fiber with no additional equipment or repeaters > >Bill > >------------------------------------------- >Bill St Arnaud >Director Network Projects >CANARIE >bill.st.arnaud(a)canarie.ca >http://www.canarie.ca/bstarn > > > > > > > >> -----Original Message----- >> From: steven hessing [mailto:stevenh@inet.unisource.nl] >> Sent: Friday, May 29, 1998 5:08 PM >> To: Bill St. Arnaud >> Cc: Jerry Scharf; nanog(a)merit.edu >> Subject: Re: flat vs non-flat charging >> >> >> Bill, >> >> WDM may lower the price per Mbit/s considerably, however many of these >> Mbit/s will have to be sold to be profitable as a service. Connecting >> two routers using dedicated WDM gear and 5km of fiber to send over >> 155Mbit/s traffic is more expensive then connecting the two routers >> directly using 5km of fiber, even though the first solution would >> give you much more available capacity. >> >> If multiple services need to run over the same fiber then existing >> ATM or SDH networks are well capable of multiplexing these services >> over one fiber at often lower costs then installing new WDM gear. >> >> There where there is a fiber shortage because there is no further >> room for service multiplexing or there is a shortage of capacity, WDM >> will be a very good solution. It is much cheaper then installing new >> fiber. >> >> WDM is therefor very real for connecting TEX's together and for >> connecting LEX's to TEX's. I expect that WDM needs OADMs or other >> up and coming technology for SDH or WDM to become cost-effective for >> the MAN/local loop. >> >> To return to the original topic, for the forseeable future (3-5 years?) >> bandwidth will remain scarce and therefor expensive. The prices per >> MBit/s will drop shortply but this will be compensated in a large >> increase in demand. An distance-based pricing scheme could therefor >> be an appropriate solution. Personally, I don't think it will happen. >> >> -- Steven >> >> PS: if anyone can prove me wrong with some numbers, please show me :) >> I've been working on cost calculations for bandwidth for the last >> couple of weeks. >> >> In your mail from 29-5-1998 you write: >> > Jerry: >> > >> > You are absolutely right. I have pulled together some numbers on public >> > pricing for long haul WDM (including right of way) showing that >> if you ou= >> > t >> > IP over WDM the cost of bandwidth drops by a factor of 100 to 1000. On >> > local loops it is more dramatic - with new WDM technology from companies >> > like Cambrian and Ciena gigabit ethernet or OC-48 SONET on a >> 5km local lo= >> > op >> > should cost about $2500 per YEAR. >> > >> > A number of our regional networks have already pulled their own >> fiber and >> > our realizing these costs in the real world. >> > >> > Fibre - $4 to $6 per meter >> > 20 year economic life and 10% maintenance per year >> > 48 strands NZDSF >> > Right of way $0 -$10 per meter per year (up to $200 per meter >> on long hau= >> > l) >> > but >> > common solution is to offer right way owner free use of a couple of >> > strands instead of paying cash >> > Installation $25 per meter underground in cities >> > $6 per meter on poles (maintenance costs 20% year) >> > Twenty year amortized cost with 10% cost of money plus maintenance plus >> > right of way costs at $10/year per meter >> > $17 per year per meter for 48 strands >> > $1.50 per year per meter per strand >> > $.50 per year per meter per wavelength >> > A 5 km OC-48 local loop should cost about $2500/year >> > >> > >> > Bill >> > >> > ------------------------------------------- >> > Bill St Arnaud >> > Director Network Projects >> > CANARIE >> > bill.st.arnaud(a)canarie.ca >> > http://www.canarie.ca/bstarn >> > >> > =A0 >> > >> > =A0 >> > >> > >> > >> > > -----Original Message----- >> > > From: owner-nanog(a)merit.edu [mailto:owner-nanog@merit.edu]On Behalf Of >> > > Jerry Scharf >> > > Sent: Friday, May 29, 1998 12:07 PM >> > > To: nanog(a)merit.edu >> > > Subject: flat vs non-flat charging >> > > >> > > >> > > I think the idea of distance charging is going away in many >> > > cases. With WDM, >> > > the cost of the WDM and SONET eqiupment on the ends of a fully >> > > populated 32 >> > > channel per fiber, 144 strand pull vastly outweight end-to-end >> > > fiber costs of >> > > anything pulled through the ground. When you add routers and the >> > > like on top >> > > of that, the distance issue really goes away and it becomes >> on of netwo= >> > rk >> > > topology hops. Can anyone with figures for new >> intercontinental pulls s= >> > ay >> > > whether this is true there as well (project oxygen marketing >> claims thi= >> > s, >> > > but...)? >> > > >> > > Using archaic telephone pricing models to argue cost of >> providing bulk = >> > IP >> > > services is just not right. >> > > >> > > jerry >> > > >> > > >> > >> >

2 1