[RC5] What's up with the stats

David McNett nugget at slacker.com
Wed Nov 19 11:08:34 EST 1997


TwinTowers says:
>This simply highlights the failure of Bovine in its capacity planning! 
>Even now that we have moved to a new home at a large ISP, I can forsee 
>very soon new problems will arise. Now we are handling data of 256 times 
>larger than RC5-56 in theory. Furthermore new features have been added to 
>the new stats. Now we have done only a friction of the keyspace donbe, 
>hence very small data. What happens a month or two down the road? Are we 
>going to wait until things are broken then only we fix them? 

A couple of minor points:
  
o It's not an issue of capacity planning.  I did not expect the 128K
   ISDN link to be sufficient at all.  It just happens to be all I have
   to work with.  Once it became a critical issue, we fixed it.  Most
   (if not all) the past, current, and future problems surrounding the
   stats server have been/are/will be simply capacity issues, as you've
   said.  Fortunately, capacity issues are typically the easiest type
   of problem to solve.  Insufficient bandwidth?  Throw a handful of T3's
   at it.  Huge logs?  Throw a handful of SCSI drives into the fray.
   Granted, it's not always that simple, but if I could have all my
   problems confined to mere capacity issues, I'd be a happy man.
   
o "very small data" is inaccurate.  Since we were able to effectively
   pick up at the peak RC5-56 keyrate very quickly, the RC5-64 stats
   database is (as of today) the same number of blocks as RC5-56 was
   on 16-Aug or so (almost six months of RC5-56 activity).  Also, since
   we've changed the teams management, we're having to track well over
   three times the number of emails as we did with RC5-56, so the
   database itself is actually much larger now than RC5-56 was in October.

And a couple of major points:

o Although Seth Dillingham broke the news (to little fanfare) last week,
   I had intended to hold off an official announcement until everything
   was finalized.  (I was also wanting the official announcement to come
   from Adam, too, but that's unimportant...)  As it stands, there are
   still a few hoops left through which we must jump before everything
   is carved in stone.  That being said, "distributed.net" is well on
   its way to becoming a recognized non-profit, tax-exempt corporation
   under 501(c)(3).  This, hopefully, will change our situation
   drastically.  Donations and gifts made to distributed.net will be
   fully deductible and this should assist us in the funding of future
   projects.  Our dream is to fund projects of interest through
   charitable donations.  We'd like, for instance, to be able to
   offer a "purse" for the completion of the next Mersenne Prime.
   Once we've sucked RSA Labs dry by completing all the RC5 challenges :)
   we'll have to look elsewhere for funds.

o In case you've not heard, _we got the money_.  Yesterday I received the
   $10,000 cheque from RSA Labs for the correct solution to the RC5-56 
   competition.  (morbidly curious readers may pause here and browse
   http://www.neosoft.com/daniel/cheque.html for more details).
   Now is the perfect time to make plans for your junket to Belgium.
   Jo's official "$1000 will buy a lot of beer" party should be happening
   pretty soon!  Also, if everything goes as planned, Duncan and I (and
   whoever else cares to attend) will be meeting in Urbana, IL on 1-Dec
   to present the $8,000 cheque to Project Gutenberg.  Although it would
   be perfectly acceptable to just mail them the money, we're all quite
   excited about this and felt that making a bit of an event out of the
   process would be fun for all involved.  Here's to this $8K donation
   being only the first of many.

>Another thing is reliance of stats on you alone. I am not trying to say 
>you are not up to the job, but the job may be too overwhelming for you. 
>Take in some assistants. Last time the stats for RC5-56 was down for a 
>long time because you were held up with your fulltime job. This time the 
>stats does not get updated because your ping time is 2000ms! What is 
>this? I thought we are into distributed business here??? Beside you, 
>can't we have someone else able to do the updating or even coding for the 
>stats?

Well, this is actually being done now.  I'm grooming a "co-stats-god"
right now who will be able to assist in the development process.  He's 
putting the finishing touches on the code to do DNS resolution of the hosts
table for the day (soon, I hope) when we bring back host statistics. There's
always a steep learning curve, however, for customized code.  And, as you
may have guessed, there's not much documentation.  :)  Already the "graphs"
portion of the stats has been effectively delegated to Dave Avery.  (and
he's doing a bang-up job at it, I might add)  Just because I'm the most
vocal (which is sometimes good, sometimes bad) doesn't mean I'm not the
only person losing sleep over the stats server.

>For those who have been discrediting the importance of stats, I must tell 
>you that whatever happens to stats adds up to the overall impression 
>about the working of Bovine: it is a closed group of "privileged" people.

The core group of individuals that could be called "distributed.net" proper
is an ever-changing (and as of late, growing) group of unique and talented
individuals.  Not a week goes by that we don't bring at least one new person
in to assist and contribute.  Yesterday, we added another person to the
OS/2 development team which makes the possibility for an OS/2 personal proxy
quite feasible.  While I admit that you're 100% dead-on correct that there
is a common misconception that we are a closed group I cannot state strongly
enough how baseless and untrue this perception is.  If you have any advice
on how we can rectify this misconception, I am eager to hear your thoughts.  
Obviously what we're doing now is not working.

>Lastly, to your credit, stats for RC5-64 has been much better than that 
>in RC5-56 in terms of features. (And not forgetting, if you didn't make 
>those many promises which you can't keep, things may indeed be better.)

I appreciate the kind words.  My own perception of distributed.net is that
we (and here I am speaking of you, I, and everyone else participating on 
any level) are the test bed.  We are the pioneers who are testing the
boundaries of distributed computing.  Although the technical difficulties
involved are non-trivial, the organizational and management issues as well
are things that need to be worked out as well.  We are the ones who have
taken on the task of testing these boundaries.  Part of that process 
involves finding out what doesn't work.  While distributed.net now, in
its infancy, tends to be a bit chaotic and seemingly disorganized, we
are building towards the day when distributed computing technologies are
a viable and efficient solution for identifying, attacking, and
accomplishing tasks that would otherwise be far beyond our individual
capabilities.

Anyway, enough ranting.  I need to get back to the stats server...  There are
plenty more promises I need to keep...  :)
-/\/ugget
 ________________________________________________________________________
|David McNett      |To ensure privacy and data integrity this message has|
|nugget at slacker.com|been encrypted using dual rounds of ROT-13 encryption|
|Birmingham, AL USA|Please encrypt all important correspondence with PGP!|

--
To unsubcribe, send 'unsubscribe rc5' to majordomo at llamas.net
rc5-digest subscribers replace rc5 with rc5-digest



More information about the rc5 mailing list