Oops, forgot about plain text formatting. I hope this is easier to read.
Quantiles (Definition 5)
Level Quantile
100% Max 5.02530E+04
99% 3.60866E+02
95% 3.89331E+01
90% 1.60667E+01
75% Q3 2.26616E+00
50% Median 1.99991E-01
25% Q1 6.62619E-02
10% 6.62619E-02
5% 6.62619E-02
1% 0.00000E+00
0% Min 0.00000E+00
Ron
-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf
Of Ron hawkins
Sent: Sunday, January 14, 2018 1:28 AM
To: IBM-MAIN@LISTSERV.UA.EDU
Subject: Re: [IBM-MAIN] AW: Re: Number of Cylinders per Volume
Radoslaw,
I'm not sure I follow. I replied to your example that specifically talked to
datasets of 50cyl allocated in CMA. I assume you chose 50 rather than 10,
because 10 is the default BPV.
If your shop has a preponderance of datasets between 10 and 50 cyls then change
the BPV so they are track managed. Then the waste is 0% and it addresses your
example exactly. You would find this skew in the univariate output. In my
sample data, the 95th percentile is 38.9 cyl.
Quantiles (Definition 5)
Level
Quantile
100% Max
5.02530E+04
99%
3.60866E+02
95%
3.89331E+01
90%
1.60667E+01
75% Q3
2.26616E+00
50% Median
1.99991E-01
25% Q1
6.62619E-02
10%
6.62619E-02
5%
6.62619E-02
1%
0.00000E+00
0% Min
0.00000E+00
The creation of 200,000 small datasets, where I think you mean less than 1
track, is not an EAV problem. You're talking about200,000 tracks of space per
day, or 42 2290-54 per year. 200K tracks is less than the track managed area of
a single volume so I don't see the problem. There would have to be an overpaid
storage admin for these datasets to end up in the cylinder managed area.
Even fragmentation can't stop you allocating trk(1) in the TMA. There are
better options with less waste suggested by Mike and Gil, and I've seen exact
matches to this other problem you described with Control/D solved with SDSP.
Ron
-----Original Message-----
From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf
Of R.S.
Sent: Saturday, January 13, 2018 12:20 PM
To: IBM-MAIN@LISTSERV.UA.EDU
Subject: Re: [IBM-MAIN] AW: Re: Number of Cylinders per Volume
Ron,
While your solution can keep "small" datasets away from EAS, it is not a
solution for a need I described.
I mean a scenarion when many thousands of relatively small datasets need to be
kept on disk. Colleague of mine administer a system, where ~200
000 datasets are created during EOD.
Of course when we talk about big storage needs it's usually not-so-many
datasets, but big or huge ones.
Last, but not least, we have limited choice - up to 64(-256) non-EAV volumes or
bigger EAV volumes (with the same limit for quantity).
Regards
--
Radoslaw Skorupka
Lodz, Poland
W dniu 2018-01-13 o 19:02, Ron hawkins pisze:
Radoslaw,
That's why you can change the threshold for the track allocation area.
DCOLLECT and PROC UNIVARIATE will probably give you all the guidance you need.
DCOLLECT does not have the primary space allocation size so this code is not
100% accurate. It is good enough to give some guidance.
I executed this against one of our general purpose lab systems, and this is the
CMA waste estimate with break point value at 5, 10 and 20 cylinders.
EAV Breakpointvalue 10 Cylinders
Total Used Cylinders 7516066.07911661
Total Wasted Cylinders 610787.607878009
Percent Wasted CMA 8.1%
EAV Breakpointvalue 5 Cylinders
Total Used Cylinders 7516066.07911661
Total Wasted Cylinders 783635.165880735
Percent Wasted CMA 10.4%
EAV Breakpointvalue 20 Cylinders
Total Used Cylinders 7516066.07911661
Total Wasted Cylinders 309321.636811143
Percent Wasted CMA 4.1%
You could use SAS code like this to come up with a good starting point for
your shop when you go EAV.
options obs=max bufno=256 nosource2 nomacrogen nosymbolgen nomlogic
nomprint sortsize=6G msglevel=i;
goptions reset=all;
%let BPV=20;
Title1 "EAV Breakpoint Value &BPV Cylinders"; *** Main
Output Title ;
FILENAME DCOLLECT FTP (
"MY.DCOL'"
)
USER='myuid' HOST='111.222.333.444' DEBUG
S370VS RCMD='SITE RDW' LRECL=32760
PASS='mypw';
proc datasets lib=pdb kill noprint;
run;quit;
proc datasets lib=work kill noprint;
run;quit;
%include sourclib(typedcol);
run;
data dcolsp;
set dcoldset;
usedcyl=dcdusesp/(15*56664);
if usedcyl>&BPV then wastecyl=mod(usedcyl,21);
output;return;
run;
title2 "Data set size distribution in Cylinders";
proc univariate data=dcolsp;
var usedcyl;
histogram;
run;
title2 "Cylinder Managed Area - Data set size distribution in
Cylinders";
proc univariate data=dcolsp;
where usedcyl>&bpv;
var usedcyl;
histogram;
run;
proc summary data=dcolsp nway;
var usedcyl wastecyl;
output out=dcolsum
sum=
;
run;
data _null_;
set dcolsum;
pctwaste=wastecyl/usedcyl;
put "EAV Breakpointvalue &BPV Cylinders"
/ "Total Used Cylinders " usedcyl
best16.
/ "Total Wasted Cylinders " wastecyl
best16.
/ "Percent Wasted CMA
" pctwaste percent16.1
;
run;
-----Original Message-----
From: IBM Mainframe Discussion List [ <mailto:IBM-MAIN@LISTSERV.UA.EDU>
mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of R.S.
Sent: Friday, January 12, 2018 5:48 AM
To: <mailto:IBM-MAIN@LISTSERV.UA.EDU> IBM-MAIN@LISTSERV.UA.EDU
Subject: Re: [IBM-MAIN] AW: Re: Number of Cylinders per Volume
W dniu 2018-01-12 o 13:15, Tom Marchant pisze:
On Fri, 12 Jan 2018 01:32:07 +0100, R.S. wrote:
BTW: an "awful waste of storage" is MCU on EAS which is 21 cyl. (315
trk). It is much more than 4kB and more than 4MB.
Not such a big deal for very large data sets, which are the most
appropriate ones to go into the cylinder-managed space. When you have
a quarter of a million cylinders, allocating space on increments of
21
cylinders isn't so bad, IMO.
Of course. However sometimes people need a lot of not-so-big datasets.
For ~1000 cyl datasets average lost is ~1% ( (21/2)/1000 ), but for 50 cyl
datasets is ~20%.
--
Radoslaw Skorupka
Lodz, Poland
======================================================================
--
Treść tej wiadomości może zawierać informacje prawnie chronione Banku
przeznaczone wyłącznie do użytku służbowego adresata. Odbiorcą może być jedynie
jej adresat z wyłączeniem dostępu osób trzecich. Jeżeli nie jesteś adresatem
niniejszej wiadomości lub pracownikiem upoważnionym do jej przekazania
adresatowi, informujemy, że jej rozpowszechnianie, kopiowanie, rozprowadzanie
lub inne działanie o podobnym charakterze jest prawnie zabronione i może być
karalne. Jeżeli otrzymałeś tę wiadomość omyłkowo, prosimy niezwłocznie
zawiadomić nadawcę wysyłając odpowiedź oraz trwale usunąć tę wiadomość
włączając w to wszelkie jej kopie wydrukowane lub zapisane na dysku.
This e-mail may contain legally privileged information of the Bank and is
intended solely for business use of the addressee. This e-mail may only be
received by the addressee and may not be disclosed to any third parties. If you
are not the intended addressee of this e-mail or the employee authorized to
forward it to the addressee, be advised that any dissemination, copying,
distribution or any other similar activity is legally prohibited and may be
punishable. If you received this e-mail by mistake please advise the sender
immediately by using the reply facility in your e-mail software and delete
permanently this e-mail including any copies of it either printed or saved to
hard drive.
mBank S.A. z siedzibą w Warszawie, ul. Senatorska 18, 00-950 Warszawa,
<http://www.mBank.pl> www.mBank.pl, e-mail: <mailto:kont...@mbank.plsąd>
kont...@mbank.plsąd Rejonowy dla m. st. Warszawy XII Wydział Gospodarczy Krajowego Rejestru
Sądowego, nr rejestru przedsiębiorców KRS 0000025237, NIP: 526-021-50-88. Według stanu na
dzień 01.01.2016 r. kapitał zakładowy mBanku S.A. (w całości wpłacony) wynosi 168.955.696
złotych.
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to <mailto:lists...@listserv.ua.edu> lists...@listserv.ua.edu with
the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions, send email to
lists...@listserv.ua.edu with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
.
