[RC5] Stats suggestion

bwilson at fers.com bwilson at fers.com
Thu Jan 4 09:47:57 EST 2001


Hi, Stephen.

We've talked about doing that very thing.  A 30- or 60-day running average
would discourage people from saving up and "megaflushing", which would be
safer for the network.

The problem we face is a technical one:  Much of the speed increase I've
attained in the rewrite of stats came from eliminating the need to
reference the huge "historical work" table.  Instead, the total work and
total rate are calculated from a running total.  The problem with an N-day
running average is that it requires a running total, and it requires work
to "fall off the end" of the table.  That means referencing the
"historical work" table for N days ago, or possibly for days
(today-N...today).  It's likely to gum up the works.  I can think of three
ways to implement this:

1)  In addition to other stats, calculate the sum(today-N...today) and
divide by N.  Problem:  Heavy use of the historical table.
2)  Keep a running-total field.  Add in "today" work, then subtract out
(today-N) work (if any).  Problem:  Still requires accessing the
historical work table, but minimizes the amount of data to be retrieved,
and localizes it to a single historical day.
3)  Maintain a separate table containing work from (today-N...today).  Add
today-work in as with the "historical work" table, and delete anything
older than (today - N).  Problem:  Disk space, adds to the complexity of
recovering from problems in stats, adds significantly to the complexity of
handling retires, team joins, etc.

I think option (2) is probably the safest choice, but it's not really an
option until we get SBIII into production.  We have lots of "priorities"
for stats, too, so it's hard to say how soon we could implement this.  We
do consider that statistic an attractive one, both from our perspective
and for the users, so it's likely to happen at some point.
__
Bruce Wilson, Manager, FERS Business Services
bwilson at fers.com, 312.245.1750, http://www.fers.com/
PGP KeyID: 5430B995, http://www.lasthome.net/~bwilson/
"A good programmer is someone who looks
both ways before crossing a one-way street."




"Stephen Berg" <sberg at pangaealink.com>
Sent by: owner-rc5 at lists.distributed.net
2001-01-04 01:30
Please respond to rc5


        To:     rc5 at lists.distributed.net
        cc:
        Subject:        [RC5] Stats suggestion


Would it be possible to have the stats server figure up a processing
rate for the last 30 or 60 days?  I've been doing RC5 since it
started.  My first system was a 486DX2/66.  I'm currently running an
Athlon 800 and three other systems.  This means my current rate is a
lot higher than my overall rate.  But daily it goes up and down
depending on when my various systems fetch and flush blocks.  With a
30 or maybe a 60 day average I'd be able to see a little more
accurately how fast my systems are working these days.


                            Stephen Berg
//-    USAF Instructor  -/-  Reluctant NT User -/- Web Designer    -//
//-                 Home = sberg at mississippi.com                   -//
//-               Work = berg.stephen at keesler.af.mil               -//
//-     http://iceberg.3c0x1.com/   -/-   http://www.3c0x1.com     -//


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





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



More information about the rc5 mailing list