MAIL ME THIS POSTING!
MY E-MAIL ADDRESS:
For example: homer@lightlink.com

From: Root Administrator <homer@lightlink.com>
To: System Notices <sn-l@lightlink.com>
Subject: Disconnects II

     SYSTEM MOTICE: sn96.script

     PLEASE DO NOT DEPEND ON OUR EQUIPMENT OR BACKUPS TO PROTECT YOUR
DATA

     light.forsale and light.jobs are usenet newsgroups.

------------------------------------------------------------------------
 
01/14/97 Tuesday 11:33pm EST

     The following is a very long and highly technical analysis of our
present modem connection rates.

     Here is the bottom line.

     Numbers are presented in the form 200 %34, this means 200 events
comprising 34 percent of the total sample.

     For the period covering the 5 days from Jan 9 to tonight Jan 14,
there were 8644 total connection attempts.

      Of these 8644 attempts,

       499 %6  were NO CARRIER or failed attempts.
      7687 %89 were sucessful V42 connections with error correction.
       458 %5  were successful NORMAL connections without error correction.

      Of the 7687 successful V42 (error correcting) connections,

      6399 %83 terminated properly under user or program control.
      1288 %17 lost carrier, meaning their line was dropped or disconnected.

      Of the 458 successful NORMAL (non error correcting) connections,

      121 %26 terminated properly under user or program control.
      337 %74 lost carrier, meaning their line was dropped or disconnected.

     In shorter words about 1 out of 5 (%17) connections is
disconnected.

      That means either everyone is disconnected 1 out of 5 connections,
or 1 out of 5 people is disconnected every time they sign on.

      The truth is probably somewhere between.

      Homer

------------------------------------------------------------------------

     Technical Analysis.

     The Multi Modem Manger is a Win95 program that monitors all
activity of the modem banks.  The computer is connected to the modem
racks via an ARCNET lan.  It not only displays date about all present
connections but also keeps a running log of all 'events' that happen
to the modem system.
 
     EVENTS

     An 'event' is someone attempting to sign on, someone failing to
make the connection, someone making the connection, someone signing
off, and someone getting dropped.

     ATTEMPTS

     The software distinguishes between two types of sign ons, failed
attempts (NO CARRIER) and sucessful attempts.

     When modems try to connect they go through a handshake as they
try to establish what speed and parameters they can talk with.  The
management software knows at this point what speed the calling modem
is trying to connect with and whether it wants V42 or NORMAL mode.

     (V42 refers to an error correcting protocol that filters out
inaudible line noise.  NORMAL mode means the modem is connected
straight through, thus any line noise comes across as data.  Line
noise at slower speeds is not significant but at 14400 it begins to
make a difference, and at 28800 it is crucial to have line noise
filtered out by having V42 error correction on.)

     When modems go through their handshake, they negotiate for
whether or not the they can maintain error correction which only works
on relatively clean lines, and what kind of error correction they are
going to use.

     If the lines are too dirty, the negotiation for error correction
will fail and the modem will attempt to connect in normal mode.

     If the lines are even worse the modem will simply disconnect with
NO CARRIER.

     When the handshaking fails totally, resulting in a NO CARRIER,
the speed and V42/NORMAL data are thrown away by the modem manager
software, and all that is left is a NO CARRIER report.

     It is thus not possible to determine after the fact whether the
NO CARRIER was the result of a V42 modem that failed to negotiate
error correction, or a NORMAL modem that didn't even try.
 
     It is almost assured though that almost all of the NO CARRIER
reports are the result of failed or no error correction.

     Because the process that modems go through to negotiate error
correction is it self flakey, we recommend that people set their V42
modems to FORCE error correction which bypasses the negotiation phase.
(This is the 'required to connect' item in Win95).
 
     This increases the likelihood of getting error correction.  It
means however that if error correction can not be attained, the line
will drop rather than allow the user on in NORMAL mode.

     Normal mode can be treasonous, it gives the apparency of being on
line, but on marginal lines will not perform properly and usually end
up disconnecting you anyway some time down the road.

     Sometimes however it is more important to be on line than to have
a reliable connection, so some people prefer to allow themselvs on in
normal mode rather than suffer not being able to get on at all.

     The forcing of error correction is called RELIABLE MODE and
forbids normal mode by dropping the line if error correction can not
be maintained.

     AUTO RELIABLE MODE means the modem is allowed to choose between
error correction and normal mode, this is what gets us into trouble
on dirty lines because it chooses normal mode rather than dropping
the line.

     Putting the modem into RELIABLE MODE which forces of error
correction is done with \N2 on Rockwell based modems, &M5 on Sportster
based modems and &E2 on Multitech based modems.

     THE MODEM REPORT

     Here is the first line of the modem report showing the statistics
for modem 1.

                    |------------V42----------|---No Error Correction--|
         Total  No                      Lost                      Lost
Modem #  Conn Carrier  Conns     OK    Carrier  Conns     OK     Carrier
-------------------------------------------------------------------------------
1-1A  1  159 | 4 %3 | 148 %93  123 %83  25 %17|  7 %4    5 %71   2 %29 |

     MODEM RACKS

     The modems are divided into 2 racks of 48 modems each.  Each rack
contains 16 cards of 3 modems each or 48 modems.

     Modem 1-1A is Rack 1, Modem card 1, modem A.  The next modem is
modem 1-1B etc.

     The number 1 to the right of 1-1A means this is modem 1.

     Then it shows 159 total connection attempts during the time
period of the report.

     In the No Carrier column it shows that there were 4 NO CARRIER
reports or 3 percent of the total connection attempts.  As explained
above here is no remaining data on whether these were V42 or NORMAL
mode attempts.

     V42 SUCCESSFUL CONNECTIONS and NORMAL SUCCESSFUL CONNECTIONS

     If the modem negotiation process succeeds this means that a
carrier has been established between the modems at some speed and
error correction (or not).  Since NORMAL mode connections are much
more likely to self destruct, it is important to separate out the V42
successful connections from the NORMAL mode successful connections.

                    |------------V42----------|---No Error Correction--|
         Total  No                      Lost                      Lost
Modem #  Conn Carrier  Conns     OK    Carrier  Conns     OK     Carrier
-------------------------------------------------------------------------------
1-1A  1  159 | 4 %3 | 148 %93  123 %83  25 %17|  7 %4    5 %71   2 %29 |

     Looking at the same line again, this shows that there were 148
successful V42 connections, and 7 successful normal connections.  The
148 represents 93 percent of the 159 total attempts on the far left,
and the 7 represents 4 percent of the total attempts.

     DISCONNECTIONS

     The manager software recognizes 5 different kinds of
disconnections.

     1.) Normal Disconnection.  This means that the connection
terminated properly, under program or user control.  This means the
hangup originates at the user's end, perhaps by hitting the
Disconnect button in Win 95, or even by hitting +++ on his terminal
emulator and typing ATH.

     Do not confuse a Normal Disconnection, which is a good thing, to
a Normal connection (no error correction) which is a bad thing.  This
is an unfortunate choice of words.

     2.) Loss of DTR.  DTR stands for Data Terminal Ready and is a
signal that Harmony, our terminal server, supplies to OUR modems at
our end to keep them on line.  When Harmony drops DTR on a modem, the
modem will hang up normally at our end.  This hangs up the user's
modem normally at his end too.  However the hangup originates at our
end rather than at the user end.

     This is what happens when a shell user types 'hangup' at the
Harmony prompt.

     Because both of these types of disconnection process are proper
and essentially under the user's control, they are both lumped
together in the data under OK.

                    |------------V42----------|---No Error Correction--|
         Total  No                      Lost                      Lost
Modem #  Conn Carrier  Conns     OK    Carrier  Conns     OK     Carrier
-------------------------------------------------------------------------------
1-1A  1  159 | 4 %3 | 148 %93  123 %83  25 %17|  7 %4    5 %71   2 %29 |

      We can see that of the 148 successful V42 connections, 122 %82
terminated OK under user control.

     If a connection is dropped without going through the proper
hangup sequence at either end, then the software registers this as a
LOSS OF CARRIER.  This happens if a user turns his modem off while on
line, or if the plug comes out the back.  It also happens if his line
gets dropped by external line conditions.

                    |------------V42----------|---No Error Correction--|
         Total  No                      Lost                      Lost
Modem #  Conn Carrier  Conns     OK    Carrier  Conns     OK     Carrier
-------------------------------------------------------------------------------
1-1A  1  159 | 4 %3 | 148 %93  123 %83  25 %17|  7 %4    5 %71   2 %29 |
 
     The data shows that of the 148 successful V42 connections, 25 %17
lost their carrier and were dropped.

     Under the No Error Correction side, you see that there were 7
successful connections, 5 terminated OK and 2 lost their carrier.

     ANAMOLIES

     There are a number of anamolies in the following data.

     1.) Some modems have significantly fewer total connection
attempts.  It is presently unclear why, but it may be that some people
were simply on for a long time on that modem.

     2.) Some modems have a significantly higher NO CARRIER and LOST
CARRIER rate.  This might be because the modem is bad, or the phone
line is bad, but it can also be because someone with a bad connection
has hunted around in the modem pool until he found a modem that did
better than others, and so he always signs onto that modem.  Even
though he gets dropped 50 percent of the time, he is still doing
better than on any other modem, so he stays there.
 
     Many of the modems in Rack 2 have such campers who staked out
that modem as their favorite modem.  They did this because they were
having a hard time with the random selection of modems that the hunt
group gives them, and they are still having a hard time, which
reflects in the higher bad numbers, but its better for them than
nothing.

     Of course if someone is already on your favorite modem when you
dial in, you will get bumped up to the next free modem anyway, however
the modems on Rack 2 are often free during unbusy times of the day, so
we often find people happily using their one modem of choice over and
over again.

     Also I personally tend to terrorize modems in Rack 2 a lot while
testing because they are not as heavily used, so like modem 2-8C shows
terrible numbers, but they reflact my efforts to drive the bad numbers
up any way I could.

     3.) It is possible to do broad based testing of different modem
settings.  For the duration of this report, all the odd modems in Rack
1 (modems 1 through 48) had fallback turned off (#F0), and all the
even modems had fallback turned on (#F2).

     It's tempting to say that all the odd numbered modems have lower
disconnect percentages than the even numbered ones, but its not
terribly significant and I doubt it would impress a statistician.  If
any of you disagree let me know.  Presently I have all modems set to
fallback-off (#F0) since the odd ones did seem to do better.

     One of the things I am going to try is reinstalling the older
firmware in the even modems, as we just installed the new firmware a
few weeks ago, and the disconnection rate has been soaring.  Of course
during this time, it has snowed outside and the weather has gotten
terrible.

     FORBIDDEN NORMAL MODE

     You will notice there are no normal mode connections at all on
Rack 2.  This is because for a while now I have had Rack 2 set to drop
normal mode connections.  This arbitrarily increases the NO CARRIER
readings for failed attempts because although normal mode connections
often disconnect later, they do often sign on, and so show up as a
successfull attempt rather than as a NO CARRIER.  With Rack 2, any
NORMAL mode modem or any V42 modem that can't negotiate error
correction will get a guaranteed NO CARRIER.

     NORMAL MODE CONNECTIONS

     The whole subject of whether we should allow normal mode
connections on our system is an interesting political problem.

     When you allow normal mode connections, a greater percentage of
people can get on, particularly those with early 14,400's which didn't
have error correction, and those with cheap 28,800's which couldn't
negotiate error correction with their own mother.

     However when normal mode connections do get on, they have a
terrible time, they have slow download speeds, flakey connections,
starts and stops, and finally they get disconnected anyhow.  
 
     First you suffer, then you die.

     Normal mode disconnections are rampantly high, around 80 percent!
 
     Or maybe a fortunate customer has clean phone lines and their
normal mode connection works for a long time, and then one day the
lines change, perhaps a mouse takes a leak on the punch down block, or
builds a nest in the phone demarc (true story), and suddenly it stops
working.  They see that 'nothing has changed' at their end, and so
they blame it on us.  Five hours of tech support later, they finally
call in the Nynex man to evict the mouse.

     14,400's without error correction are a 50/50 bet at best on
Ithaca lines trying to connect to us.

     Even if they start off working 'fine', over time they are bound
to fail as line quality comes and goes with the season's and the hand
of God.

     Normal mode connections are therefore a technical nightmare.  In
particular people blame the lousy quality of their connection on your
server or your modems or you or anything but their own modem.

     FORBIDDING NORMAL MODE

     If we forbid normal mode connections by setting OUR modems to
refuse them, we find out right up front if someone doesn't have a good
modem, or has a cheap V42 that can't negotiate.  You then either set
the V42 to force error correction, or tell them to get a new modem.

     If you do this though, those who are too poor to upgrade, or
those stuck at 2400 which work fine in normal mode, can't get on.  We
have a lot of 2400 baud users.

     THEY AREN'T THE PROBLEM.  IT'S THE CHEAP 14,400's and THE LOUSIER
28,800's THAT ARE CAUSING ALL THE TROUBLE.

     COST SHIFTING

     On the other hand, if we allow normal mode mode connections, we
find ourselves pressured to purchase more and more expensive modems
like the new USR Couriers ($6000) just so everyone else can get away
with buying cheap $50 Win Modems that should never be allowed on the
store shelves.
 
     The USR Couriers and Sportster by the way are both based on the
same technology, so the fact that our Couriers make it easier for the
cheap 28,800's to connect indicates that our users would be wise to
get a USR Courier or Sportser!  But really the higher end Rockwell's
and Multitech's are fine too.  Higher end though usually means $130
and above.  Not $85 or below.

     Allowing normal mode connections allows our users to get away
with an insideous form of cost shifting, from them to us.  We have to
buy more expensive modems so they can buy cheaper modems.

     I am willing to shell out the $6000 for the USR Couriers, but I
am very tempted to set them to reject normal mode connections.

     That way those with cheap modems will be forced to upgrade, and
then they will do fine on the Multitechs leaving the Couriers for
those with truely difficult lines.

     But locking out normal mode on any modem also locks out the
business man who just *HAS* get a piece of e-mail from his account,
even if he has to try 10 times because he is calling from an oil rig
in the middle of the atlantic ocean during a hurricane using a cell
phone and an old accoustic coupler modem.  (True story.) With normal
mode locked out, he wouldn't stand a chance.

     Also sometimes people with very high quality modems simply live
in an area of town where the lines are so poor they can do nothing
about them.  They have the best modems, but still can't get on AT ALL,
unless we let them use normal mode even on the Couriers.  Forbidding
normal mode connections would lock out these people too.

     So it's a problem.  In general the 14,400's are phasing out, and
almost everyone is moving up to 28,800's now, which *ALL* have V42
error correction, but the cheap ones still fail to negoatiate it on
rough lines without having it forced on.  And even then some of them
don't make it.

     THE DATA

     So here is the data.  If you see trends that we might be
interested in, please let us know.  I will be posting this to our web
site periodically, so we can keep track of changes in the situation as
the weather changes and as I make experimental changes in our settings
and arrangements.

 
     Jan 9, 1997 - Jan 14 1997

                    |------------V42----------|---No Error Correction--|
         Total  No                      Lost                      Lost
Modem #  Conn Carrier  Conns     OK    Carrier  Conns     OK     Carrier
-------------------------------------------------------------------------------
1-1A  1  159 | 4 %3 | 148 %93  123 %83  25 %17|  7 %4    5 %71   2 %29 |
1-1B  2  225 | 3 %1 | 216 %96  186 %86  30 %14|  6 %3    1 %17   5 %83 |
1-1C  3  132 | 3 %2 | 126 %95  110 %87  16 %13|  3 %2      %0    3 %100|
1-2A  4  277 |10 %4 | 248 %90  203 %82  45 %18| 19 %7    6 %32  13 %68 |
1-2B  5  220 | 5 %2 | 204 %93  164 %80  40 %20| 11 %5    4 %36   7 %64 |
1-2C  6  211 | 8 %4 | 186 %88  151 %81  35 %19| 17 %8    4 %24  13 %76 |
1-3A  7  235 | 9 %4 | 214 %91  175 %82  39 %18| 12 %5    2 %17  10 %83 |
1-3B  8  203 |11 %5 | 173 %85  142 %82  31 %18| 19 %9    2 %11  17 %89 |
1-3C  9  189 | 8 %4 | 164 %87  141 %86  23 %14| 17 %9    4 %24  13 %76 |
1-4A  10 153 |11 %7 | 129 %84   97 %75  32 %25| 13 %8    2 %15  11 %85 |
1-4B  11 210 | 9 %4 | 185 %88  155 %84  30 %16| 16 %8    3 %19  13 %81 |
1-4C  12 218 |13 %6 | 189 %87  149 %79  40 %21| 16 %7    3 %19  13 %81 |
1-5A  13 175 |12 %7 | 153 %87  133 %87  20 %13| 10 %6    3 %30   7 %70 |
1-5B  14 105 | 3 %3 |  92 %88   75 %82  17 %18| 10 %10   3 %30   7 %70 |
1-5C  15 161 | 3 %2 | 150 %93  125 %83  25 %17|  8 %5    1 %13   7 %88 |
1-6A  16 177 | 6 %3 | 165 %93  142 %86  23 %14|  6 %3    1 %17   5 %83 |
1-6B  17 150 | 6 %4 | 137 %91  111 %81  26 %19|  7 %5    1 %14   6 %86 |
1-6C  18 161 | 5 %3 | 146 %91  124 %85  22 %15| 10 %6    1 %10   9 %90 |
1-7A  19 132 | 6 %5 | 119 %90  101 %85  18 %15|  7 %5    1 %14   6 %86 |
1-7B  20 137 | 4 %3 | 122 %89  107 %88  15 %12| 11 %8    3 %27   8 %73 |
1-7C  21 191 | 8 %4 | 166 %87  139 %84  27 %16| 17 %9    5 %29  12 %71 |
1-8A  22 109 | 6 %6 |  96 %88   78 %81  18 %19|  7 %6    2 %29   5 %71 |
1-8B  23 168 | 4 %2 | 148 %88  129 %87  19 %13| 16 %10   2 %13  14 %88 |
1-8C  24 171 | 8 %5 | 158 %92  124 %78  34 %22|  5 %3    1 %20   4 %80 |
1-9A  25 200 |14 %7 | 169 %85  140 %83  29 %17| 17 %9    5 %29  12 %71 |
1-9B  26 152 |11 %7 | 134 %88  113 %84  21 %16|  7 %5    6 %86   1 %14 |
1-9C  27 201 |12 %6 | 174 %87  151 %87  23 %13| 15 %7    7 %47   8 %53 |
1-10A 28 144 | 5 %3 | 128 %89  111 %87  17 %13| 11 %8    2 %18   9 %82 |
1-10B 29  83 | 4 %5 |  71 %86   63 %89   8 %11|  8 %10   3 %38   5 %63 |
1-10C 30 113 | 7 %6 |  98 %87   91 %93   7 %7 |  8 %7    1 %13   7 %88 |
1-11A 31 125 |13 %10| 105 %84   91 %87  14 %13|  7 %6    3 %43   4 %57 |
1-11B 32 156 |11 %7 | 135 %87  113 %84  22 %16| 10 %6    5 %50   5 %50 |
1-11C 33 122 | 1 %1 | 111 %91  100 %90  11 %10| 10 %8    5 %50   5 %50 |
1-12A 34 110 | 2 %2 | 103 %94   85 %83  18 %17|  5 %5    1 %20   4 %80 |
1-12B 35 159 | 9 %6 | 141 %89  119 %84  22 %16|  9 %6      %0    9 %100|
1-12C 36 119 | 5 %4 | 103 %87   89 %86  14 %14| 11 %9    2 %18   9 %82 |
1-13A 37 128 | 8 %6 | 112 %88   93 %83  19 %17|  8 %6    2 %25   6 %75 |
1-13B 38 118 | 3 %3 | 108 %92   87 %81  21 %19|  7 %6    2 %29   5 %71 |
1-13C 39 105 | 3 %3 |  97 %92   81 %84  16 %16|  5 %5    3 %60   2 %40 |
1-14A 40 118 | 5 %4 | 110 %93   93 %85  17 %15|  3 %3    1 %33   2 %67 |
1-14B 41 117 | 7 %6 | 104 %89   89 %86  15 %14|  6 %5    2 %33   4 %67 |
1-14C 42 128 | 8 %6 | 112 %88   91 %81  21 %19|  8 %6    1 %13   7 %88 |
1-15A 43 120 | 8 %7 | 106 %88   83 %78  23 %22|  6 %5    2 %33   4 %67 |
1-15B 44 111 | 8 %7 |  98 %88   84 %86  14 %14|  5 %5    1 %20   4 %80 |
1-15C 45  99 | 8 %8 |  85 %86   66 %78  19 %22|  6 %6    2 %33   4 %67 |
1-16A 46  93 | 4 %4 |  87 %94   72 %83  15 %17|  2 %2    1 %50   1 %50 |
1-16B 47 100 | 7 %7 |  82 %82   71 %87  11 %13| 11 %11   4 %36   7 %64 |
1-16C 48  76 | 7 %9 |  66 %87   56 %85  10 %15|  3 %4      %0    3 %100|
2-1A  49 159 |10 %6 | 149 %94  129 %87  20 %13|    %0      %0      %0  |
2-1B  50 123 | 6 %5 | 117 %95  105 %90  12 %10|    %0      %0      %0  |
2-1C  51 136 |11 %8 | 125 %92  104 %83  21 %17|    %0      %0      %0  |
2-2A  52 103 | 7 %7 |  96 %93   80 %83  16 %17|    %0      %0      %0  |
2-2B  53 111 | 4 %4 | 107 %96   91 %85  16 %15|    %0      %0      %0  |
2-2C  54  87 | 5 %6 |  82 %94   69 %84  13 %16|    %0      %0      %0  |
2-3A  55  59 | 3 %5 |  56 %95   35 %63  21 %38|    %0      %0      %0  |
2-3B  56  80 |15 %19|  65 %81   57 %88   8 %12|    %0      %0      %0  |
2-3C  57  64 | 5 %8 |  59 %92   48 %81  11 %19|    %0      %0      %0  |
2-4A  58  37 | 5 %14|  32 %86   29 %91   3 %9 |    %0      %0      %0  |
2-4B  59  43 | 8 %19|  35 %81   30 %86   5 %14|    %0      %0      %0  |
2-4C  60  39 | 3 %8 |  36 %92   27 %75   9 %25|    %0      %0      %0  |
2-5A  61  16 | 1 %6 |  15 %94    2 %13  13 %87|    %0      %0      %0  |
2-5B  62  36 |   %0 |  36 %100  32 %89   4 %11|    %0      %0      %0  |
2-5C  63  36 | 7 %19|  29 %81   24 %83   5 %17|    %0      %0      %0  |
2-6A  64  27 | 6 %22|  21 %78   15 %71   6 %29|    %0      %0      %0  |
2-6B  65  49 | 5 %10|  44 %90   41 %93   3 %7 |    %0      %0      %0  |
2-6C  66  26 |10 %38|  16 %62   10 %63   6 %38|    %0      %0      %0  |
2-7A  67  27 | 8 %30|  19 %70   11 %58   8 %42|    %0      %0      %0  |
2-7B  68  24 |11 %46|  13 %54    8 %62   5 %38|    %0      %0      %0  |
2-7C  69   1 |   %0 |   1 %100     %0    1 %10|    %0      %0      %0  |
2-8A  70  45 |16 %36|  29 %64   17 %59  12 %41|    %0      %0      %0  |
2-8B  71  27 | 9 %33|  18 %67    9 %50   9 %50|    %0      %0      %0  |
2-8C  72  23 | 9 %39|  14 %61   10 %71   4 %29|    %0      %0      %0  |

------------------------------------------------------------------------
SUMMARY OF MODEM CONNECTION DATA
------------------------------------------------------------------------

Total Connection Attempts:    8644

Total No Carrier:              499 %6

|----------------------V42-----------------------|
    Good                                  Lost
 Connections           OK                Carrier
  7687 %89           6399 %83            1288 %17

|-------------No Error Correction----------------|
    Good                                  Lost
 Connections           OK                Carrier
   458 %5             121 %26             337 %74

 
              Loss of Carrier by Time of Day

            |------V42------|   |----NORMAL----|
            Total    Losses     Total    Losses
Time        Losses   During     Losses   During
------------------------------------------------------------------------
00:00:00    1288      14 %1     337       13 %4
01:00:00    1288      16 %1     337        6 %2
02:00:00    1288      21 %2     337        8 %2
03:00:00    1288       9 %1     337        1 %0
04:00:00    1288       4 %0     337          %0
05:00:00    1288      14 %1     337        6 %2
06:00:00    1288      24 %2     337        1 %0
07:00:00    1288      47 %4     337       11 %3
08:00:00    1288      74 %6     337       19 %6
09:00:00    1288      82 %6     337       10 %3
10:00:00    1288      82 %6     337       14 %4
11:00:00    1288      81 %6     337       12 %4
12:00:00    1288      69 %5     337       24 %7
13:00:00    1288      75 %6     337       26 %8
14:00:00    1288      87 %7     337       18 %5
15:00:00    1288      96 %7     337       24 %7
16:00:00    1288      78 %6     337       15 %4
17:00:00    1288      62 %5     337       17 %5
18:00:00    1288      63 %5     337       10 %3
19:00:00    1288      47 %4     337       13 %4
20:00:00    1288      43 %3     337       21 %6
21:00:00    1288      97 %8     337       20 %6
22:00:00    1288      72 %6     337       31 %9
23:00:00    1288      31 %2     337       17 %5