December 2003 - cypherpunks-legacy

Re: Emergency Broadcast System
by cme＠ellisun.sw.stratus.com 17 Dec '03

17 Dec '03

>Message-Id: <9310221941.AA16004(a)toad.com> >Subject: Emergency Broadcast System >Date: Fri, 22 Oct 1993 13:36:33 -0600 (MDT) >From: "Philippe_D_Nave" <pdn(a)dwroll.dw.att.com> > 1) Mailing list data should be distributed to "N" sites where "N" is a > magic number that minimizes the chance of losing all the copies. > (Mathematicians, sharpen your pencils!) I found the equations I mentioned earlier. The following is a latex file giving some. The trend should be obvious. You don't need many nodes, if you can repair broken ones quickly enough. [This is why Stratus has such a large FedEx bill. When one of our machines gets a broken component, it phones home automatically, reports the failure and a new part is FedEx'ed immediately. If the report comes in by midnight, the new part is on the customer's desk for him/her to replace by 9am. (Our current machines have N=2, K=1). We prefer high speed mailaing of parts (from a warehouse at FedEx's central sorting airport) over normal mail in order to maximize $\mu$.] ------------------------------------------------------------------------------ \documentstyle[12pt]{article} \begin{document} \title{MTTF of Various Systems} \author{Carl M. Ellison \thanks{Stratus Computer Inc., 55 Fairbanks Blvd., Marlborough MA 01752. Email address: {\tt cme(a)sw.stratus.com}.}} \date{June 18, 1993} \maketitle \begin{abstract} Equations are presented for the Mean Time To Failure (MTTF) of various systems, as a function of the number of nodes in the system, N, and the minimum number of nodes in a working system, K. Failure of a system is defined as having fewer than K working nodes. \end{abstract} \section{Equations} Here are some equations for the Mean Time To Failure (MTTF) of various systems, as a function of the number of nodes in the system, N, and the minimum number of nodes in a working system, K. Failure of a system is defined as having fewer than K working nodes. Typically, K=1 and each node has a complete copy of each database. However, sometimes the data can be kept on multple nodes (as in a RAID-5 disk array) which will tolerate some failures, down to a given threshold. It is assumed that as soon as a failure occurs, a repair cycle will be started. There is then a race to see if the repair can be completed before enough additional nodes fail to drop the working number below K. If that race is lost, it is assumed that information has been lost, in the worst case, but at least that the service represented by the N nodes is not available. These equations do not give system availability (probability that a service is available) but rather MTTF. Availability equations can be found in several good textbooks. See, for example, Trivedi's text on Statistics. MTTF is more difficult to compute. The equations below were each a symbolic solution to a custom Markov chain, built to model the choice of N and K. N: number of nodes in a full system K: number of nodes in a minimally functional system $\lambda$: rate of failures (e.g., number of node failures per year) $\mu$: rate of node repair (in the same units as $\lambda$) Each fraction below is the MTTF of the whole system: the mean time until a system drops to only (K-1) working nodes. N = 3 ; K = 2 \begin{equation} \frac{ 5\lambda + \mu }{ 6\lambda^2 } \end{equation} N = 4 ; K = 2 \begin{equation} \frac{ 26\lambda^2 + 6\lambda\mu + \mu^2 }{ 24\lambda^3 } \end{equation} N = 4 ; K = 3 \begin{equation} \frac{ 7\lambda + \mu }{ 12\lambda^2 } \end{equation} N = 5 ; K = 2 \begin{equation} \frac{ 154\lambda^3 + 36\lambda^2\mu + 7\lambda\mu^2 + \mu^3 }{ 120\lambda^4 } \end{equation} N = 5 ; K = 3 \begin{equation} \frac{ 47\lambda^2 + 8\lambda\mu + \mu^2 }{ 60\lambda^3 } \end{equation} N = 5 ; K = 4 \begin{equation} \frac{ 9\lambda + \mu }{ 20\lambda^2 } \end{equation} N = 6 ; K = 2 \begin{equation} \frac{ 1044\lambda^4 + 240\lambda^3\mu + 48\lambda^2\mu^2 + 8\lambda\mu^3 + \mu^4 }{ 720\lambda^5 } \end{equation} N = 6 ; K = 3 \begin{equation} \frac{ 342\lambda^3 + 60\lambda^2\mu + 9\lambda\mu^2 + \mu^3 }{ 360\lambda^4 } \end{equation} N = 6 ; K = 4 \begin{equation} \frac{ 74\lambda^2 + 10\lambda\mu + \mu^2 }{ 120\lambda^3 } \end{equation} N = 6 ; K = 5 \begin{equation} \frac{ 11\lambda + \mu }{ 30\lambda^2 } \end{equation} N = 7 ; K = 2 \begin{equation} \frac{ 8028\lambda^5 + 1800\lambda^4\mu + 360\lambda^3\mu^2 + 62\lambda^2\mu^3 + 9\lambda\mu^4 + \mu^5 }{ 5040\lambda^6 } \end{equation} N = 7 ; K = 3 \begin{equation} \frac{ 2754\lambda^4 + 480\lambda^3\mu + 75\lambda^2\mu^2 + 10\lambda\mu^3 + \mu^4 }{ 2520\lambda^5 } \end{equation} N = 7 ; K = 4 \begin{equation} \frac{ 638\lambda^3 + 90\lambda^2\mu + 11\lambda\mu^2 + \mu^3 }{ 840\lambda^4 } \end{equation} N = 7 ; K = 5 \begin{equation} \frac{ 107\lambda^2 + 12\lambda\mu + \mu^2 }{ 210\lambda^3 } \end{equation} N = 7 ; K = 6 \begin{equation} \frac{ 13\lambda + \mu }{ 42\lambda^2 } \end{equation} N = 8 ; K = 2 \begin{equation} \frac{ 69264\lambda^6 + 15120\lambda^5\mu + 3000\lambda^4\mu^2 + 520\lambda^3\mu^3 + 78\lambda^2\mu^4 + 10\lambda\mu^5 + \mu^6 }{ 40320\lambda^7 } \end{equation} N = 8 ; K = 3 \begin{equation} \frac{ 24552\lambda^5 + 4200\lambda^4\mu + 660\lambda^3\mu^2 + 92\lambda^2\mu^3 + 11\lambda\mu^4 + \mu^5 }{ 20160\lambda^6 } \end{equation} N = 8 ; K = 4 \begin{equation} \frac{ 5944\lambda^4 + 840\lambda^3\mu + 108\lambda^2\mu^2 + 12\lambda\mu^3 + \mu^4 }{ 6720\lambda^5 } \end{equation} N = 8 ; K = 5 \begin{equation} \frac{ 1066\lambda^3 + 126\lambda^2\mu + 13\lambda\mu^2 + \mu^3 }{ 1680\lambda^4 } \end{equation} N = 8 ; K = 6 \begin{equation} \frac{ 146\lambda^2 + 14\lambda\mu + \mu^2 }{ 336\lambda^3 } \end{equation} N = 8 ; K = 7 \begin{equation} \frac{ 15\lambda + \mu }{ 56\lambda^2 } \end{equation} N = 9 ; K = 3 \begin{equation} \frac{ 241128\lambda^6 + 40320\lambda^5\mu + 6300\lambda^4\mu^2 + 888\lambda^3\mu^3 + 111\lambda^2\mu^4 + 12\lambda\mu^5 + \mu^6 }{ 181440\lambda^7 } \end{equation} N = 9 ; K = 4 \begin{equation} \frac{ 60216\lambda^5 + 8400\lambda^4\mu + 1092\lambda^3\mu^2 + 128\lambda^2\mu^3 + 13\lambda\mu^4 + \mu^5 }{ 60480\lambda^6 } \end{equation} N = 9 ; K = 5 \begin{equation} \frac{ 11274\lambda^4 + 1344\lambda^3\mu + 147\lambda^2\mu^2 + 14\lambda\mu^3 + \mu^4 }{ 15120\lambda^5 } \end{equation} N = 9 ; K = 6 \begin{equation} \frac{ 1650\lambda^3 + 168\lambda^2\mu + 15\lambda\mu^2 + \mu^3 }{ 3024\lambda^4 } \end{equation} N = 9 ; K = 7 \begin{equation} \frac{ 191\lambda^2 + 16\lambda\mu + \mu^2 }{ 504\lambda^3 } \end{equation} N = 9 ; K = 8 \begin{equation} \frac{ 17\lambda + \mu }{ 72\lambda^2 } \end{equation} N = 10 ; K = 4 \begin{equation} \frac{ 662640\lambda^6 + 90720\lambda^5\mu + 11760\lambda^4\mu^2 + 1400\lambda^3\mu^3 + 150\lambda^2\mu^4 + 14\lambda\mu^5 + \mu^6 }{ 604800\lambda^7 } \end{equation} N = 10 ; K = 5 \begin{equation} \frac{ 127860\lambda^5 + 15120\lambda^4\mu + 1680\lambda^3\mu^2 + 170\lambda^2\mu^3 + 15\lambda\mu^4 + \mu^5 }{ 151200\lambda^6 } \end{equation} N = 10 ; K = 6 \begin{equation} \frac{ 19524\lambda^4 + 2016\lambda^3\mu + 192\lambda^2\mu^2 + 16\lambda\mu^3 + \mu^4 }{ 30240\lambda^5 } \end{equation} N = 10 ; K = 7 \begin{equation} \frac{ 2414\lambda^3 + 216\lambda^2\mu + 17\lambda\mu^2 + \mu^3 }{ 5040\lambda^4 } \end{equation} N = 10 ; K = 8 \begin{equation} \frac{ 242\lambda^2 + 18\lambda\mu + \mu^2 }{ 720\lambda^3 } \end{equation} N = 10 ; K = 9 \begin{equation} \frac{ 19\lambda + \mu }{ 90\lambda^2 } \end{equation} \end{document}

1 0

MTTF expressions for K = 1
by cme＠ellisun.sw.stratus.com 17 Dec '03

17 Dec '03

Sorry -- I left out the K=1 expressions from that paper. They're a little more mundane (for Stratus) but of obvious interest here. These are the most likely to be interesting to us. Consider, for example, \lambda = 0.5 failures/year and \mu = 52 repairs/year for the kind of machines we're talking about (as opposed to Stratus machines). The resulting MTTF would be in years. - Carl ==================================================================== N = 2 ; K = 1 \begin{equation} \frac { 3\lambda + 1\mu }{ 2\lambda^2 } \end{equation} N = 3 ; K = 1 \begin{equation} \frac { 11\lambda^2 + 4\lambda\mu + 1\mu^2 }{ 6\lambda^3 } \end{equation} N = 4 ; K = 1 \begin{equation} \frac { 50\lambda^3 + 18\lambda^2\mu + 5\lambda\mu^2 + 1\mu^3 }{ 24\lambda^4 } \end{equation}

1 0

Digital signatures to be required?
by greg＠ideath.goldenbear.com 17 Dec '03

17 Dec '03

My C-punks feed comes over a UUCP connection, and I only poll once or twice per day, under normal circumstances. I'd likely never notice a 6-hour delay, and hence wouldn't give a shit one way or the other. Delaying message posting seems likely to exacerbate the "ask a question, get the same answer 8 times" problem. If we're going to reward some behaviors and punish others, I'd prefer to see something that'd create exponentially increasing delays for > 1 post per person per 24 hours; of course, remailers make that impractical. Imposing a digital signature requirement strikes me as an arbitrary exercise of authority which would be largely without benefit. I doubt that people interested in the C-punks list are the folks who need to be dragged kicking & screaming into the 21st century; at least not in terms of personal use of technology. :) Count mine as a "NO" vote; let's talk about the delay thing again when we come up with a filter that can spot "what is [anarchy|libertarianism] anyway?" or bogus pseudo-economic jibberish. -- Greg Broiles greg(a)goldenbear.com Baked, not fried.

1 0

Re: pseudospoofing survey
by an41418＠anon.penet.fi 17 Dec '03

17 Dec '03

I don't know the value of such a survey. I am not inclined to answer these questions to someone who I don't really know over a communication system that is inherently untrustworthy. I agree that this information would be valuable could it be collected, but I figure since I don't feel like revealing this information, especially not under my true identity, then others, (who may even be malicious at times) will be less likely to. What is your intention in posting such questions? Do you really expect to learn anything from the responses? Wonderer ------------------------------------------------------------------------- To find out more about the anon service, send mail to help(a)anon.penet.fi. Due to the double-blind, any mail replies to this message will be anonymized, and an anonymous id will be allocated automatically. You have been warned. Please report any problems, inappropriate use etc. to admin(a)anon.penet.fi.

1 0

bye for now...
by swc＠uc1.ucsu.edu 17 Dec '03

17 Dec '03

The last couple of weeks have been interesting and fun (and have provided me with useful information upoin which I will act shortly), but I can't keep up with the volume. Thanks, and see you later (when I have more time). I will remain on the -announce list and will still be available via direct e-mail. bye for now... ------------------------------------------------------------------------ Stuart W. Card, Consultant, Card & Associates -- Research & Development Box 153 RR 1 Newport Rd Utica NY 13502 315-735-1717 / FAX -8469 swc(a)uc1.ucsu.edu or cards(a)top.cis.syr.edu "Who is John Galt?"

1 0

ADMIN: proposed new policy on the mailing list
by nobody＠soda.berkeley.edu 17 Dec '03

17 Dec '03

-----BEGIN PGP SIGNED MESSAGE----- I don't think the security problems of Eric's proposal are as bad as some have suggested. People could create a separate, "lower security," public key for use on multi-user systems, with a different pass phrase than for their "high security" key. Perhaps the key could have an appropriately small size of about 400 bits. It would still be far more work to break the security of such a system than to forge mail across the network (which is easy). So such a system does enhance security over the existing system of unsigned messages. As for the argument that people don't have access to signature systems - PGP illegal in the U.S., PEM/RIPEM unavailable outside the U.S. - just turn these around: PGP legal and available outside the U.S., PEM/RIPEM legal and available inside the U.S. Everyone has access to legal encryption and signature software. One problem I see with Eric's suggestion is that it is couched in terms which suggest our main business here is debate. Eric suggests that a delay in having your message appear is to your disadvantage because your opponent's arguments will stand unrefuted for a time. Recently the list has been pretty contentious, but historically there has been much less debate here than on many other lists, and I would hope that we could return to that approach. It would be better if we could exchange information, ideas, approaches for reaching the goals we share. In such an environment a delay in the posting of a good idea is to everyone's detriment and does not particularly harm the person whose ideas were slow to appear. Still, I support Eric's basic goal of encouraging more use of the technologies we talk so much about. I will remind people that I make an encrypted version of the cypherpunks list available to anyone who has a PGP public key which includes their address. There are only a few sub- scribers now but if you'd like to try it send me your PGP public key and I'll add you to the list. I'd suggest trying it for a few days before cancelling your regular CP subscription to make sure you can handle the encrypted traffic. Ironically, our anonymous posters, who have generated so much controversy of late, are at least using the technology. Maybe if we do implement some form of Eric's idea we should give the preferential treatment to anonymous posters as well as signed messages. People can post anonymously without having to expose any secret information, which should address the security concerns mentioned above. I'm sending this message with the following headers, which direct replies back to my address. This reduces the biggest inconvenience with using remailers, the inability to get replies: :: Request-Remailing-To: cypherpunks(a)toad.com ## Reply-To: hfinney(a)shell.portal.com One more point. I was the one who initially implemented the clear-sign feature in PGP. I copied the idea from PEM, but put the signature info at the end. (I still think that putting it at the beginning was a silly idea.) When I wrote it I didn't have the blank lines around the ----- separators. Branko added those and also add the Version: line (which is ignored by the software). I still like my format better, as I think the signatures should add as little as possible. But have you seen what these PEM signatures look like? I've seen a couple on sci.crypt and I almost fell out of the chair laughing. First, the guy had to _manually_move_ the signature from the beginning to the end to get it out of the way. But, worse, the signature is like thirty or forty lines long! I kid you not. The guy posts a message of about a (24-line) screenful then it has like two screenfuls of signature information. It looks ludicrous. No wonder he had to move the signature to the end - otherwise people would have given up before they even got to his message. (In fairness, these PEM signatures are self-checking; PGP signatures require you to get the key on your own. I'm not sure if a non-self-checking PEM signature mode exists.) Hal Finney hfinney(a)shell.portal.com -----BEGIN PGP SIGNATURE----- Version: 2.3a iQCVAgUBLMnCI6gTA69YIUw3AQHs2AP5Ae64wUfiNa4/yborffvMry1MAt9chF05 9Bdz3NupXkWU1GNbmniFKDnU+GdGR+Tuu3HgwwV7N55EjLY7SclOaLBxKXySD25X sAlwlH1yDZO/ly5UxKakdaPKR4nzIZZjPZ8ZoCkDszoNcxERj/nF7l7zLYP3eXF+ GG+YBHenSL4= =/09p -----END PGP SIGNATURE-----

1 0

Re: Mail delivery question
by peter honeyman 17 Dec '03

17 Dec '03

well, i think bill overstates the case. but then, i'm from the slash/hack, no risk/no gain school of computer programming. peter

2 1

Proposed policy: look like sig
by gwt＠eskimo.com 17 Dec '03

17 Dec '03

Eric said, >Given the objections I've received, I now amend my proposal from "sign >your messages, or else" to "make something that looks like a >signature, or else". This has several consequences that I >particularly like. Wouldn't it "look like a signature" if I just copied at the end of an article the actual signature of some other, signed, article? That would be a true signature, just not mine and not for this article. It sounds as though it would meet the suggested requirement, but I don't see how it would accomplish the goals. I could do it with a macro, or maybe use a special .sig for posts to this list. It wouldn't add anything to my software. What am I missing? -George

1 0

Re: impersonation
by an5877＠anon.penet.fi 17 Dec '03

17 Dec '03

-----BEGIN PGP SIGNED MESSAGE----- Scott, I'm not sure what you mean by dissemination of information not being ergodic -- could you please elaborate. Well, not elaborate, because your style is already quite elaborate, and I can't make heads or tails out of your note. The comparison of Email forgery to rape ("or any other invasion of privacy" -- excuse me, but forcibly shoving your penis into my vagina is not an invasion of my privacy, it is a violent crime on my body, for which you will pay) so enrages me, it makes calm discussion impossible. DEADBEAT <na5877(a)anon.penet.fi> -----BEGIN PGP SIGNATURE----- Version: 2.3 iQBFAgUBLMrfAPFZTpBW/B35AQHTvAGAnV21LCA7m5AeD05EoAUbjuLAHTZK9lHM Y1C/1B2n6wmYAivNaCAbvk+YJHza5Hv5 =wh4r -----END PGP SIGNATURE----- ------------------------------------------------------------------------- To find out more about the anon service, send mail to help(a)anon.penet.fi. Due to the double-blind, any mail replies to this message will be anonymized, and an anonymous id will be allocated automatically. You have been warned. Please report any problems, inappropriate use etc. to admin(a)anon.penet.fi.

1 0

Viallage Voice article, etc
by nate＠VIS.ColoState.EDU 17 Dec '03

17 Dec '03

Does anyone have a copy of the Viallage Voice article from the 3rd of August 1993? I don't want to type it in, and I know I saw it on the list a while back. Also, I am compiling a list of articles that explain what we are all about (I know, there is no CypherPunks agenda per se, but there are common goals here), to be given out to the public. If you are in posession of such an article (or know of one, I'll type it in) send it to me, please. Thanks, -nate -- +-----------------------------------------------------------------------+ | Nate Sammons email: nate(a)VIS.ColoState.Edu | Colorado State University Computer Visualization Laboratory | Finger nate(a)monet.VIS.ColoState.Edu for my PGP key | #include <std.disclaimer> Guerrilla Cryptographer | Title 18 USC 2511 and 18 USC 2703 Protected --> Monitoring Forbidden +--------+ Always remember "Brazil"

1 0