- Test - lists.nanog.org

Re: Comparison of provider's network status access
by Sean Donelan 31 May '00

31 May '00

On Tue, 30 May 2000, jlewis(a)lewis.org wrote: > On 26 May 2000, Sean Donelan wrote: > > > Sprint - Has a web site describing scheduled and emergency maintenance > > mailing lists. Scheduled maintenance mailing list has some > > messages, but I've never seen a message about an outage on the > > emergency maintenance list. > > I get almost (sometimes more than) daily outage notices on > outage(a)sprint.net. IIRC, they had a link to subscribe on their web site. I subscribed, my sprint salesperson verified I subscribed, I also subscribed to the DALnet copy of the sprint list. No notices. No one at sprint seemed very interested in figuring out why, and now I'm no longer a Sprint customer I'm not very interested now either.

2 1

Re: Shopping for NOCs
by mike harrison 31 May '00

31 May '00

> Simple question: What do you look in a NOC when considering someone > for service? In other words, what would be the ideal NOC? Forget technology, facilities and everything but the quality of the people involved. --Mike--

2 1

Comparison of provider's network status access
by Sean Donelan 31 May '00

31 May '00

On Fri, 26 May 2000, Martin Cooper wrote: > INSnet (admittedly not a tier-1) has a pretty neat policy on > disclosure, even including a detailed description of the fixes > for problems as well as the causes of them, which is nice. Essentially all of the top, tier-2, tier-3, etc providers have some type of online network status web site. How well they use it to keep users of their network informed varies greatly. Abovenet - Web site with MRTG graphs, and copies of messages from their tech contact mailing list about problems, outages, and scheduled maintenance. AGIS - no more. Closed access to their network status web site years ago. AOL - I don't know, I haven't found anything on their public web site. AT&T - Network status web site with red, yellow, green city to city matrix. Outage portion of the web site is never updated. Worldnet only newsgroups frequently updated with detailed problem reports. C&W - Traffic web site (red, yellow, green), but I've noticed they seem to only have "green" dots on the front map. You have to dig to find the problem sites. Has an outage mailing list, but it never has any mail and even C&W employees seem unsure of its purpose. Exodus - BENGI web site with major trouble tickets available. Genuity/GTEI/BBN - customer only network status web site. "Finger" access to trouble tickets, including master tickets. Good proactive notification of customers of problems. Globalcenter - customer only web site. Mindspring - Mindspring has a public network status web site with current trouble tickets. PSI - used to have a detailed "finger" and web site network status page, starting last month I get an access denied message. SBC/PBI - Web site with current network status. Frequently updated and includes resolution or reason for outage. Sprint - Has a web site describing scheduled and emergency maintenance mailing lists. Scheduled maintenance mailing list has some messages, but I've never seen a message about an outage on the emergency maintenance list. UUNET - web site updated for some major outages. Says details will follow, but usually the entire web page is "refreshed" to all systems normal message every few hours whether or not the problem is fixed. Some customers reportly received mail messages about outages.

2 1

Re: pop server in an ISP environment
by Dave Crocker 30 May '00

30 May '00

At 06:37 PM 5/30/00 +0800, Muljawan Hendrianto wrote: > XTND XMIT, may I know what is that ? It is a non-standard extension to POP, supported in some POP client and server implementations. It provides an authenticated alternative to open SMTP posting. The current standard for authenticated posting is the AUTH command, coming into support on clients and servers. Note that email posting is now a separate standard (SUBMIT) from email relaying (SMTP) event though their core protocol engine is currently the same. d/ =-=-=-=-= Dave Crocker <dcrocker(a)brandenburg.com> Brandenburg Consulting <www.brandenburg.com> Tel: +1.408.246.8253, Fax: +1.408.273.6464 675 Spruce Drive, Sunnyvale, CA 94086 USA

2 3

Re: NANOG Registration Request
by Carol Wadsworth 30 May '00

30 May '00

For those of you who have registered to attend NANOG 19 in Albuquerque, June 11-13, your registration fee must be received by June 2, otherwise the fee will increase $50. To check on the status of your payment, please see the meeting attendee list: http://www.nanog.org/mtg-0006/Registration.List.html If there are dashes by your name, payment is still due. ------------------------------ Payment Submission Information ------------------------------ Registration Fee: Before June 2 $300 After June 2 $350 Cancellation Fee (after 6/2) $35 Credit Card: When paying by credit card--MasterCard, Visa, or American Express--please include the following information: Card Type Number Expiration Cardholder's Name Checks: Please make payable to the University of Michigan Mail or fax payment information to: NANOG Registrar Merit Network, Inc. 4251 Plymouth Road, Suite 2000 Ann Arbor, MI 48105-2785 Fax Number: (734) 647-3185 --On Tue, Mar 28, 2000 10:39 AM -0800 Automated NANOG Registrar <NANOG(a)merit.edu> wrote: > ser(a)tch.org (Steve Rubin) submitted the following > NANOG Registration: > --------------------------------------------------------- > --- Name: Steve Rubin > Organization: IP Networks, Inc. > Demographic: Network Operator > Profit Status: For Profit > E-Mail Address: ser(a)tch.org > > Payment Method: Will Send Check > Credit Card #: No CC info given > > Server protocol: HTTP/1.0 > Remote host: 199.74.220.2 > Remote IP address: 199.74.220.2 > >

1 0

Re: NANOG Registration Request
by Carol Wadsworth 30 May '00

30 May '00

For those of you who have registered to attend NANOG 19 in Albuquerque, June 11-13, your registration fee must be received by June 2, otherwise the fee will increase $50. To check on the status of your payment, please see the meeting attendee list: http://www.nanog.org/mtg-0006/Registration.List.html If there are dashes by your name, payment is still due. ------------------------------ Payment Submission Information ------------------------------ Registration Fee: Before June 2 $300 After June 2 $350 Cancellation Fee (after 6/2) $35 Credit Card: When paying by credit card--MasterCard, Visa, or American Express--please include the following information: Card Type Number Expiration Cardholder's Name Checks: Please make payable to the University of Michigan Mail or fax payment information to: NANOG Registrar Merit Network, Inc. 4251 Plymouth Road, Suite 2000 Ann Arbor, MI 48105-2785 Fax Number: (734) 647-3185 --On Tue, Mar 28, 2000 10:39 AM -0800 Automated NANOG Registrar <NANOG(a)merit.edu> wrote: > fsd(a)psi.com (Frank DiGennaro) submitted the following > NANOG Registration: > --------------------------------------------------------- > --- Name: Frank DiGennaro > Organization: PSINet > Demographic: Network Operator > Profit Status: For Profit > E-Mail Address: fsd(a)psi.com > > Payment Method: Will Send Check > Credit Card #: No CC info given > > Server protocol: HTTP/1.1 > Remote host: 206.102.68.79 > Remote IP address: 206.102.68.79 > >

1 0

Re: pop server in an ISP environment
by Dave Crocker 30 May '00

30 May '00

Muljawan Hendrianto, Selamat tenga hari. I forgot to include some references: At 06:37 PM 5/30/00 +0800, Muljawan Hendrianto wrote: > XTND XMIT, may I know what is that ? It is a non-standard extension to POP, supported in some POP client and server implementations. It provides an authenticated alternative to open SMTP posting. The current standard for authenticated posting is the AUTH command, coming into support on clients and servers. Note that email posting is now a separate standard (SUBMIT) from email relaying (SMTP) event though their core protocol engine is currently the same. References: RFC 2476 Message Submission. R. Gellens, J. Klensin. December 1998. (Format: TXT=30050 bytes) (Status: PROPOSED STANDARD) RFC 2554 SMTP Service Extension for Authentication. J. Myers. March 1999. (Format: TXT=20534 bytes) (Status: PROPOSED STANDARD) d/ =-=-=-=-= Dave Crocker <dcrocker(a)brandenburg.com> Brandenburg Consulting <www.brandenburg.com> Tel: +1.408.246.8253, Fax: +1.408.273.6464 675 Spruce Drive, Sunnyvale, CA 94086 USA

1 0

Re: pop server in an ISP environment
by Peter van Dijk 30 May '00

30 May '00

On Tue, May 30, 2000 at 11:22:21AM -0400, Alex Rubenstein wrote: > > > Indeed. For example, for the company I work for (that is burdened by the > > 64k limit right now because of lot's of stock linux 2.0.x kernels), I am > > developing a mail-solution that doesn't chew up UIDs for popboxes. > > Try the user-less qmail install, HOWTO at: > > http://www.tibus.net/pgregg/projects/qmail/single-uid-howto.html > > We're about to implement this. I have looked at it but it does not fit my needs at all. The system I'm developing (which _is_ based on qmail tho), has several advantages. Also, I don't need 'user-less' stuff, because all customers already have a hosting-account for FTP and shell. The big advantage is that users own and control their own mailinglists, popoboxes, forwards and all that and normally don't need _any_ intervention from us to get anything done. This also allows for easy quota-enforcing and accounting - all stuff from one customer is under 1 UID. With that user-less stuff, AFAICS the admin needs to do work for each popbox created. This ofcourse can be automated, but is still a centralized process. We are moving towards decentralization as far as possible while still keeping a grip. Greetz, Peter. -- petervd(a)vuurwerk.nl - Peter van Dijk [student:developer:madly in love]

1 0

RE: pop server in an ISP environment
by Roeland M.J. Meyer 30 May '00

30 May '00

> Muljawan Hendrianto: Tuesday, May 30, 2000 3:38 AM > > thanks to everybody who has replied to my email ! > Yes, my sole concern is actually the performance of my pop > server when my pop3 users base getting larger. > But I think with a E250/220 running having 2 CPUs , 1MB of > RAM, around 200GB HD I should be able to handle up to 20000 > pop3 users easily, shouldn't I ? Are you sure that you have only 1MB RAM, or is that a typo? Did you mean to say 1GB? > Roeland, you mentioned in your email about XTND XMIT, may I > know what is that ? It is the means where a client can send mail via the POP3 channel, rather than SMTP. Some clients can do this, if it is available. We have tested Netscape messenger and Eudora for this feature. There are some serious benefits to this, like being able to lock clients into a single POP host, forbidding them to see other SMTP hosts, or ANY SMTP hosts. Thus, keeping all email internal connections, internal. It is part of en extended POP specification that is part of the POP3 RFCs. Do a search through the IETF database for XTND XMIT. Only two servers implement this however, qpopper is one of them. Also, look in the qpopper source code. It is fully documented there. That's where I found it the first time. > And about HA for Qpopper, are you saying that you have HA > agent for Qpopper ? It is much more complex than that. We wrote a lot of code. But, we are also a software development house. We did a fully distributed architecture that apportioned users over a large number of hosts, allowing us to keep the user load limited, dependent on host capacity and expected load. I have heuristics that calculate that. It is a front-end application system, similar to what AOL has to run.

1 0

pop server in an ISP environment
by Muljawan Hendrianto 29 May '00

29 May '00

Hello all, I would like to have your opinions regarding pop server set-up in an ISP environment, what would be the common software used, authentication type etc. I am thinking about using QPOPPER+procmail, but some people say that it is not scalable because its authentication is based on /etc/passwd. And in Unix environment there is certain recommendation not to have more than 5000 users in /etc/passwd file. Another issue is high availability, does any body use server clustering in an ISP environment? I am thinking about having a Sun Cluster for these pop servers, but will I need a special HA agent ? AFAIK, the qpopper daemon will be initiated by the users (via inetd) when they are accessing their mailbox. Hence a Sun cluster will not need a special agent because, if one of the servers crashed, the clustering software should manage to redirect any traffic to the second server, so that the users will connect transparently to that qpopper daemon in the second server. Does clustering in the real world can do the above scenario? What about LDAP for user's authentication ? Is it recommended to use ? Any suggestions/hints will be greatly appreciated. ps: the system should be able to accomodate about 20000 users. regards, Muljawan

8 7