May 2001 - NANOG - lists.nanog.org

Re: heads up on 80/8 [actually operational, so don't read this!!!]
by Travis Pugh 17 May '01

17 May '01

Ah yes ... I was hoping for some email notification that is sent out when RIPE breaks open a new block, since I really don't check the iana.org assignments page on a regular basis ... something akin to what ARIN does w/north american allocations. -travis On Thu, 17 May 2001, Antony wrote: > > This info is there hidden on ripe ncc website > a complete source of info can be found at. > > http://www.ripe.net/ripencc/mem-services/general/allocs.html > Yes google will find this if you search specifically. > > I assume apnic should update their page soon. > http://www.apnic.net/db/RIRs.html > > RIR allocations can also be found at, > http://www.iana.org/assignments/ipv4-address-space > > > -antony > > > On Thu, May 17, 2001 at 10:49:28AM -0400, Travis Pugh wrote: > > > > > > I can't for the life of me find anything on the web about RIPE recieving > > or allocating from 80/7 ... google has failed me. Can someone in RIPEland > > let some of us north americans know where announcements of new RIPE > > allocations are made? I'll happily forward any email notification to > > NANOG, as I'm sure many others would, but where does the initial > > announcement get sent? I'm sure many of us would like to update our > > prefix filters in a timely fashion. > > > > Cheers. > > >

1 0

Re: heads up on 80/8 [actually operational, so don't read this!!!]
by smd＠clock.org 17 May '01

17 May '01

| Several weeks old info. I am suprised that ICANN and or RIPE haven't made | the operational announcement to the NANOG or other lists. Not to invite a flamewar or anything, but it's not really an operational issue until some operator tries to route whatever they are allocated by the IRRs (RIPE, ARIN, APNIC). None of the IRRs (or ICANN) can really make ANY promise whatsoever that their allocations are globally routable. (This will be particularly true of ever-longer prefixes as they are allocated by various IRRs. IOW, who cares if the IRRs hand out /32s or /128s - just so long as people understand that the odds of global reachability in both steady state and during convergence [*] decrease roughly proportionally to the length of the prefix). So, this could well have been the first time an operator has found an operational issue with 80/8 that is relevant to the NANOG audience. :-) | Thanks for posting it to NANOG You're very welcome - hey, first public-act-of-smd in many days. Sean. - -- Sean Doran <smd(a)use.net> / <smd(a)ebone.net> - -- [*] http://www.ripe.net/docs/ripe-210.html

5 4

Broadband security?
by Joe Blanchard 17 May '01

17 May '01

Please excuse the redundancy regarding this topic, but I can't help publisizing this. A few weeks ago I came across a rather odd opening regarding a certain ISP and its Cayman's DSL routers. Oddly enough I found 19 open routers actually telling/publisizing that there was no password set for the admin account using little more than a web browser (of course one of them being mine). A bit concerned I contacted the ISP in question, their NOC to be exact, and told them of this. During the conversation it was pretty clear that the NOC person didn't really care, and that "The customer is responsible for that security" or better said (not my job) applied within this situation. A bit concerned I contacted a sales rep from that same ISP and got this "wow, really, can you send me those IP addys and we'll look into it right away!". With some strange sense of helpfulness I sent those items to him and heard not a word. That was about 2 weeks ago, and I again checked on the nodes I had seen them open and found the same openings. I thought perhaps it was just a honey pot, but after changing two of the routers then restarting them and seeing the changes I knew nothing had been done. On a whem, and sort of a bet, I did a scan of the ISP's net and found over 100 Cayman routers open, as well some odd 20 SpeedStream routers (simple password/login just give it admin and you have the keys to the kingdom so to speak). To me, and perhaps I am missing something here, This seems a bit odd, in that a major ISP deploying these items would in fact leave routers, ok junior routers, this wide open. I really don't want to name the ISP in question openly for the obvious reasons, but has it really gotten to the point that Broadband for businesses is slapped in with no security and no education to the persons getting it? Sorry for the rant -Joe

3 2

RE: To CAIS Engineers - WAKE UP AND TAKE CARE OF YOUR CUSTOMERS
by John Murphy 16 May '01

16 May '01

I'm sorry Roeland, but your statement about SWIP'ing anything less than a /24 is just inaccurate. We SWIP /29s and /30s daily, for both our Frame and our DSL customers. -----------------------Original Message----------------------- From: Roeland Meyer To: nanog(a)nanog.org Date: 5/13/01 4:21 PM Subject: RE: To CAIS Engineers - WAKE UP AND TAKE CARE OF YOUR CUSTOMERS I've had similar problem at SpeakEasy. They still don't have a reverse-DNS clue. http://www.mhsc.com/recovery.htm None of the DSL ISPs can do larger than /27 anymore, even when they're ILECs. Anything less than a /24 can't be SWIP'd and if you don't control your in-addr.arpa entries you don't control your domain and have no security. > From: John Palmer (NANOG Acct) [mailto:nanog@adns.net] > Sent: Sunday, May 13, 2001 1:40 PM > <FLAME ON> > PSINet dumped all of their DSL customers onto CAIS. Covad is > the backhaul > provider for them both. Need I say more? > > Bottom line, I dont know how many thousands of people were > without service > for more than a day because the whole transfer was botched. > > We have a situation where we had a DSL connection from PSI > and were using > our own IP addresses. Problem is that no one changed the > routing tables and the packets dead-ended at PSI. > > Getting CAIS to fix this problem has been a nightmare. At > first, PSI didnt > stop announcing the routes and now that they have, it seems that > CAIS will not announce the routes till Monday becuase "no one > at our NOC > knows how to do this and the one guy (ONE GUY IN THE WHOLE > COMPANY - AND THEY ARE A NATIONWIDE PROVIDER????) who knows > how doesn't work > weekends." > <flame off - sorry to the list for being so loud>

7 9

RE: The DSL business model
by Roeland Meyer 16 May '01

16 May '01

According to the SBA, most small business are 30-75 employees. > From: Jared Mauch [mailto:jared@puck.Nether.net] > Sent: Wednesday, May 16, 2001 1:27 PM > > Small business, so up to 500 employees? > > ip per user plus the IS/servers related to infrastructure, > subnetting, you can easily justify some real address space. > > - jared > > On Wed, May 16, 2001 at 04:22:10PM -0400, Shawn McMahon wrote: > > On Wed, May 16, 2001 at 03:54:07PM -0400, Daniel Golding wrote: > > > > > > There is a simple solution for redundancy - buy redundant > T-1 lines, run > > > BGP, route a large enough block of space to be globally > routable. This is > > > > Who's going to give a small business a large enough block to be > > globally routable?

1 0

RE: The DSL business model
by Roeland Meyer 16 May '01

16 May '01

> From: Steve Sobol [mailto:sjsobol@NorthShoreTechnologies.net] > Sent: Monday, May 14, 2001 2:18 PM > > [I don't know that I agree that this is off-topic...] I'm with you here. > Steve Schaefer wrote: > > > 1) ISP's who deploy DSL are on a very slim margin. They > can't afford to > > be "full service." That doesn't mean that the best DSL ISP's don't > > provide better service than the mediocre or poor DS-1 > ISP's, but don't > > expect $500/month support for your $150/month line. I disagree with this characterization. Quite a few of those businesses would gladly pay a few shekels more, for a decent SLA and some service. The problem is in how you stratify the market. There are also market aggregation issues (crossed with excessive greed). > > 2) Data CLEC's who provide most of the lines for > independent ISP's have a > > challenging business model. Northpoint went down hard. > That doesn't mean > > that the others will go down hard, too, but they have the > same kind of > > risks. They needed lots of capital and got most of it through debt > > financing. Debt financing is okay, as long as you don't overstate your sales projections. Homes/business' passed isn't [the same as] homes/business' online. Getting them online is the actual gross revenue (minus delinquencies) and you have to have your payment terms fall under some reasonable percentage of that. What many of them did was ass-u-me a larger percentage of plant utilization (ie x plant generates y sales, over z months). The sales didn't ramp up, or the delinquency rate went out of sight, and predictable results followed. It's no different with public offerings. Not making your dividend payments is about the equivalent of not making your loan payments. It's all credit. > Well, wasn't the big problem that the DLECs were having > trouble turning > around orders, and as a result ended up not getting paid by the ISPs? Maybe. Looking at the Covad v. DSLnetworks case, Covad let DSLnetworks get about 18 months in arrears. It certainly smells like bad management of receiveables. How many of you guys would let arrearages age more than 90 days, regardless of the reason. At worst, you convert it to a prommisory note, or some other debt instrument, with re-payment terms spelled out in their BoD's blood. > > 3) ILEC's don't have a clue. Some of them are well-intentioned > to say about both Verizon and Ameritech. Ameritech was > fair-to-middling > until being bought by SBC, and they rolled downhill rapidly Having witnessed SBC takover of PacBell, from the inside (PacBell ACN/CBS), it was most amusing to watch San Ramone meeting San Antone. I absolutely knew that the HFC system was going to get cut, at the minimum. SBC track-record isn't real good and they're not all that nice to work for. > Here in Cleveland, Ameritech offers only ADSL. I am not sure they even > know what SDSL is. :) They will, however, be happy to sell > you 1.5x256 ADSL for $175 a month! PacBell does NOT offer SDSL, which is pretty much a requirement for most business usage. ADSL is only good as a [barely] competitive offering against residential cable. BTW, other than DSLnetworks failure, my DSL line has been rock solid for over 2 years. Every problem I've had was an upstream failure. You can't fault xDSL technology, only the business' behind it. The technology that can be faulted, and I sincerely apologize for not being clear enough on this, is that which prevents adequate redundancy at the end-nodes. The fact that 100,000 businesses can get lopped off by a single provider business failure is pretty sad. I've tried, for over a year, to get redundant uplinks to an alternate provider (ISDN backup to xDSL). CIDR, prefix filtering, and cluelessness nail that effort every time. It doesn't seem to matter that I am more than willing to pay for it. It simply isn't available. But, it should be (I don't mean tinker-toy methods either). Before CIDR, it was. The past few quarters has shown how necessary it is. Guys, this is a huge market gap, why isn't anyone filling it? -- ROELAND M.J. MEYER Managing Director Morgan Hill Software Company, Inc. TEL: +001 925 373 3954 FAX: +001 925 373 9781 http://www.mhsc.com mailto: rmeyer(a)mhsc.com

6 6

RE: 2 Wire T1 - HDLC
by Martin, Christian 16 May '01

16 May '01

John, I believe you are referring to HDSL2, which can and does run over single-pair but uses advanced spectral shaping and error correcting mechanisms that weren't invented when HDSL was dreamed up by Bellcore. ./chris > -----Original Message----- > From: John A. Tamplin [mailto:jat@liveonthenet.com] > Sent: Monday, May 14, 2001 7:13 PM > To: Hunter Pine > Cc: mike harrison; nanog(a)nanog.org > Subject: Re: 2 Wire T1 - HDLC > > > > On Mon, 14 May 2001, Hunter Pine wrote: > > > That was my mistake Mike. HDSL is NOT two SDSL lines by > definition. It's > > just two circuits running symmetric 768k DSL (not SDSL) > bonded to form a > > single1.5mb line. > > Umm, no. HDSL rides a single pair, and provides more than 1.5Mb. > > John A. Tamplin jat(a)jaet.org > 770/436-5387 HOME 4116 Manson Ave > 770/431-9459 FAX Smyrna, GA 30082-3723 > >

2 1

Instant chats and central servers
by Sean Donelan 16 May '01

16 May '01

A question (and a test to see if I'm still subscribed) The various instant messenging services, such as AIM, ICQ, Microsoft, Yahoo, other Messenger uses a central server to manage "presence". No central server appears to mean no instant messages, am I correct? What does this have to do with NANOG, apparently it is becoming more common for backbone NOC folks to communicate with their friends in other NOCs via one of these instant chat programs. I didn't realize how common it was until I was informed about it last month when AOL/AIM had difficulties. This month Yahoo Messenger had power difficulties, which disrupted their central servers. If folks are using this these services for real-time communications, should we be trying to improve their reliability? Or is this just a "feature" of how presence services work.

15 20

heads up on 80/8 [actually operational, so don't read this!!!]
by smd＠clock.org 16 May '01

16 May '01

Hello, long time no see - RIPE is now allocating subnets of 80/8 to end users. In order to avoid the ugliness of various people announcing 80/8 without guaranateeing transit to all the subnets, if you are filtering subnets of this /8, please relax your filters to allow /20s and shorter, On behalf of random Europeans, Sean. (yeah, ok, this is at least partially work-related-stuff...)

5 5

corrigendum - 80/7 not 80/8 - smd-is-blind
by smd＠clock.org 16 May '01

16 May '01

I was reminded by a colleague: | 80/7 to be correct even :) | | see http://www.ripe.net/ripe/mail-archives/routing-wg/current/msg00006.html | for copy of RIPE's announcement email wrt this. This is why we get professionals to do LIR stuff. :-) Sean. PS - LIR = RIPE terminology (local internet registry), and roughly means an ISP who allocates address space to others

1 0