May 2001 - Test - lists.nanog.org

Re: EMAIL != FTP
by David Lesher 26 May '01

26 May '01

> i've seen queues backing up because the bank of 1200 bps modems weren't > enough, and queues backed up because a series of 400Meg messages kept failing > halfway through delivery. Doesn't anyone here recall the infamous OSU/Compuserve ftp-by-mail debacle...??? It totally killed OSU's mail system. In a case of fiction portends reality, Karl Kleinpaste solved it by applying a Net-trueism; he wrote the backlog off to 9-track tapes and DROVE them across town.... -- A host is a host from coast to coast.................wb8foz(a)nrk.com & no one will talk to a host that's close........[v].(301) 56-LINUX Unless the host (that isn't close).........................pob 1433 is busy, hung or dead....................................20915-1433

2 1

RE: EMAIL != FTP
by Roeland Meyer 26 May '01

26 May '01

> From: Jim Mercer [mailto:jim@reptiles.org] > Sent: Friday, May 25, 2001 5:47 PM > On Fri, May 25, 2001 at 11:22:15AM -0400, deeann mikula wrote: > you are helping to create a society of users who think that > Netscape/Explorer > is the internet, with no concept of any of its other applications. You miss the point. It's what the users want. They are not interested in maximizing efficiency <gasp>. They are into what works. They don't give a flip how. > > unfortunately, i have accepted the fact that there is no stopping > > customers from emailing 10MB (and larger!) files, they can > just go to > > our competition if we prevent it. > > that would be my tactic, but then i'm my own boss, so i can make those > decisions. > > i get the impression that you (or someone else on the > "pro-ftp-by-email" side) > is "just and employee" and has to do as they are told. When I started my own company, the first time, I was under the nieve impression that I was my own boss and could do whatever I wanted. Guess what? I wound up with more bosses then you could shake a stick at. They're called CUSTOMERS. EVERYONE has a boss! If you don't, then you are already richer than Gates or broke as hell. Fifteen years later (and lots of peaks and valley's) have not changed my mind on that. Owning your own company just makes you feel the pain that much more directly. > well, have a bit of guts, and try to explain to management what the > implications (both current and local as well as future and > far-reaching) of such policies are. Everyone learns to pick their battles. This issue is so far down the priority list with ALL of my clients that you'd need a Saturn V booster, just to find the bottom of the well. Since they pay me to know better, they would simply question my sanity ... and proceed to find some one that will do the job they wanted to get done. > when 100 customers leave because 1 customer hosed your mail > server, then maybe > they will start listening to tachnical advice, as opposed to > relying solely > on marketing advice. The smart ones find a way to have the cake AND eat it.

1 0

RE: Email Parcel Post is Good!
by owen＠dixon.delong.sj.ca.us 26 May '01

26 May '01

> ... and you expect to get this disk from Livermore to Chicago, in 5 minutes, > ... how? > > Business moves at Internet speeds these days, in case you hadn't noticed. > > To put it in a way even understandable to techies; > > Collaborrative document, updated and swapped between 4 authors, and > reviewed by two others, widely separated geo-physically. Revised and > rewritten 5 times per day, for a week (Not counting NetMeetings). Trust me, > management does work this way. > Network Shared storage is the correct solution for most of these issues. Runtime with network shared storage is also 1 week. > Process run-time, with email = 1 week; with FedEx = 1 month. > > Process cost (transit only), with email <= $20, with FedEx = $4500 (FedEx = > very happy) > Let's see... 4 authors * 35 revisions (7 days 5 revisions each, or 35 days, 1 revision each, assuming revisions wouldn't be reduced by latency) = 140 fedex packages. According to FedEx web site, from 95121 (San Jose, CA) to 20170 (Herndon, VA), it's: $18.34 standard overnight (w/unscheduled pickup) $20.94 priority overnight (w/unscheduled pickup) I get $2567.60 and $2931.60, respectively. Still well over the $20 that it costs with network shared storage or email, but definitely significantly less than your fedex estimate. In fact, I couldn't get close to your estimate on the fedex web site. I was able to find 8am delivery at $6000+, but that's way more than your $4500, and not overly useful in this situation. Admittedly, I didn't try HI to VA, so maybe that's where you got your quote. OTOH, most businesses have contracts with fedex and pay substantially less than the one-time rates I just listed. > Real cost; Deal is blown out because deliverable is three weeks late. > Instant lost revenue. > > > my opinion. > > opinion: invalid. > > Fact: Using e-mail instead of bulk file transfer or network shared storage is an inefficient use of bandwidth with little or no gain in user convenience and only an artificial sense of security. Owen

1 0

Email Parcel Post is Good!
by Cutler, James R 26 May '01

26 May '01

Various people are alleged to have said things like: "file transfer by email is evil" ========================================================= Business Response: "Parcel Post" is just as valid for electronic media as it is for physical media. Of course it would be stupid, most of the time, to use email to transfer the Windows 2000 Service Pack 2 (~100 Megabytes), but it can be done. (It truth, I have never mailed an attachment larger than 50 Megabytes.) The business arguments for casual exchange of documents using email are, I believe, strong. Creation of an FTP site or a Web site is NOT a good idea for the casual user, even ignoring the difficulty of explaining how to make one work using a transient IP address. Security and support are difficult enough today without adding that complication. Arbitrarily denying the use of email for file transfer is, in and of itself, tortuous interference with legitimate business of others. - James R. Cutler - EDS 800 Tower Drive, Troy, MI 48098 1 248 265 7514 james.cutler(a)eds.com

4 3

Re: EMAIL != FTP
by Richard A. Steenbergen 25 May '01

25 May '01

On Fri, May 25, 2001 at 03:11:13PM -0400, Wojtek Zlobicki wrote: > > Last time that I checked, adding a drive to a raid array or NETAPP was > a great deal more than $99. Lets not count the cost of upgrading a HD > in a lowly Linux box (not bashing Linux here). IDE is also not > sufficient enought in order to sustaint the write rate of that these > servers demand (let alone the IO). 5200 RPM drivers are not a > solution for ISPs. Ack. I promised myself I was going to stay out of this thread but I don't think I'm going to be able to resist. If you're going to insult IDE you should at least know what you're insulting it for. Your average 60gig ($140) or 80gig ($200) 5400rpm IDE Maxtor drive will have no problems reading or writing at 30MB/s sustained. I have built software IDE raid setups that have no problem reading at 50MB/s sustained with boatloads of CPU to spare, and are not light on the seek ability especially as the number of drives increase. Your average 7200rpm IBM 75gig drive (a bit more expensive, $250/ea) will perform a lot better then that. I'm looking at an extremely loaded web server right now pushing over 90Mbit/s at over 600 hits/sec and over 400 drive seeks per second, off a 500gig raid-5 built entirely from cheap 7200rpm IDE drives. That's not your multi-terabyte mail and news server for hundreds of thousands of users, but I bet you its a lot cheaper. The only area where IDE really loses is logistics and cable management, if you're going to keep drives balanced on seperate IDE channels. With UDMA/66 controllers at $14/ea from Fry's, there is no technical reason you can't get some great performance out of IDE drives at extremely cheap prices. If you want to take it a step more towards business, look into 3Ware products for IDE drives. Don't forget what the I in RAID stands for. But enough about off-topic stuff, I don't want to distract from the next line of discussion which will be if you didn't write your email in vi you shouldn't have the right to send it. -- 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)

1 0

Re: Splitting a block of Class C's
by Albert Meyer 25 May '01

25 May '01

At 04:23 AM 5/25/01 -0400, John Fraizer wrote: >Huh? You mean they're not running their own DNS servers? If they're not >running their own DNS servers but are doing hosting, they deserve what >they get. If they are, they need to simply update their host object in >whois for their DNS servers. (This may or may not work with "alternate >registries".) Then, they can write a simple script to go through the zone >files they have and change the first three octets to the new address >space. I mentioned that, and it turns out they meant s/domains/servers/ but they really really don't want to re-IP, which I can understand. I had to re-IP 10 servers when my former employer moved offices, and it wasn't much fun. They have 50+ boxes in their space, and about half are client colos. In light of the advice I've heard here and elsewhere, I'm a bit nervous about what will happen, but I did tell them I would do it if Sprint agreed. I expected that Sprint would say no, but they claim that Sprint has agreed.

1 0

NANOG T-shirt gallery
by Sean Donelan 25 May '01

25 May '01

Has anyone created a gallery of all the NANOG T-shirt designs and put it on-line anywhere?

2 1

Re: Stealth Blocking
by dlr＠bungi.com 25 May '01

25 May '01

[In the message entitled "RE: Stealth Blocking" on May 24, Roeland Meyer <rmeyer(a)mhsc.com> writes:] > > I'm getting seriously confused here. I thought that the open-relay issue was > irelevent to MAPS. That MAPS only black-holed confirmed SPAM sites (a little > tougher, but more granular, charter). Further, that it was ORBS that listed > open-relay sites specifically, whether they were involved in a spam or not > (unacceptable due to punishing potential anti-spammers for proliferating > spam that never saw their systems). To me, these are two entirely different > charters. If MAPS starts to look like ORBS then I will stop using MAPS. > > Can someone please clarify? Sure. MAPS has four real-time lists. The MAPS RBL(sm) is a list of sites and networks which are known to be friendly, or neutral to spam. They include sites which harbour known spam origin points, multi-hop open relays which refuse to close (and have transmitted spam), spamware sites, and other persistant spam sources. Hosts and networks can use this list via DNS (rejecting mail, and other traffic as they see fit), or BGP (usually blackholing all traffic bound for those sites). It's very hard to get a site listed. It's quite easy to get off the RBL, assuming that the issue that caused the listing has been corrected. The MAPS RSS(sm) is a list of open relays *which have been abused*. These are sites which have been reported to MAPS as open relays, and have spam samples. Once the spam has been verified, a test is performed to verify that the site is, indeed, an open relay. If a sample message is accepted, and then returned by the site as a relay, the host is listed. Removal from the RSS requires that the host no longer relays. Automated probes are never done - a human must request the test, and spam must be available. Because of the very large number of hosts listed (around 100,000 as I write this), it's generally used in DNS mode only. It's pretty easy to get a host which is an open relay that has transmitted spam onto the list. Between 100 and 1,500 hosts per day are added, and hundreds per day are taken off (as soon as they let MAPS know that the relay has been closed). The MAPS DUL(sm) is a list of dialup ports. These are dialups which have been reported to MAPS by the ISP running them, or by users which have received spam from the dialup. An investigator verifies that the address range does contain dialup ports before they are listed. Hosts and networks typically use this list in DNS mode to reject direct-from-dialup spam. It's time-consuming to get an address listed, and also time-consuming to get an address removed from this list. The MAPS RBL+(sm) is a combined list, which allows a single lookup to search all lists. It's possible to use this in BGP mode, but it's unlikely that anyone would want to do so. So, does MAPS look like ORBS? ORBS probes systems no matter if spam has emitted or not. Does this catch more open relays? You bet. Is it network abuse to scan for open relays? I think so. Do spammers use the same techniques? You bet. MAPS probes systems only after they have been abused by spammers. Does this allow spammers to use the relays for at least one spam run? You bet. Is it network abuse to confirm an open relay that has transmitted spam? I don't think so. Do spammers use the same techniques? I don't think so. ORBS probes systems periodically after they are listed to see if they have been closed. Does this ensure that relays are removed after they are secured? Sort of. But this requires that the hosts listed be probed frequently, and still doesn't ensure that they are removed "as soon as they are secured". MAPS depends on the system administrator contacting MAPS to get a host de-listed. Does this ensure that they are removed once they are secured? Sort of. It does require that the admin be willing to contact MAPS. So, does MAPS look like ORBS? You decide. I'm certain that all of the network owners on this list want spam to stop. MAPS is one tool that can help. ORBS is another. Where you draw the line, or how you protect your networks is your choice. One thing that I think *will* help, particularly in the short term, is port 25 blocking of dialup ports. It's my personal opinion that this will have the greatest impact on spammers who abuse open relays. I've watched this happen over the last few months, as various large networks have secured their dialup ports. It's impressive. --

12 18

Re: PacBell Frame Relay Outage?
by Richard A. Steenbergen 25 May '01

25 May '01

On Fri, May 25, 2001 at 12:48:40PM -0700, Sean Donelan wrote: > > Pacbell statement at 11:04am PDT: > > Some customers in California and Reno may be unable to access the > Internet at this time do to an equipment problem. Our Engineering > Team is working to resolve this issue. We apologize for any > inconvenience this may cause and thank you for your patience. How delightfully non-descriptive. I hope it was at least "due" to an equipment problem. Wouldn't it be easier if they just issued a magic 8ball instead of a statement? "When will my internet be up?" "Reply hazy, try again" -- 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)

1 0

Re: PacBell Frame Relay Outage?
by Roy 25 May '01

25 May '01

> Thanks to everyone for their info and help. According to my source, the problem was in the Newbrodge switches. Tonight's outage is for maintenance on the Lucent gear. My frame circuits returned a few minutes ago but DSL still seems affected. Roy Engehausen PS.. Sorry to have pestered everyone for the operational issues :-)

1 0