Conference koolit::vms_curriculum

I will be updating the VMS Performance Management L/L and Sem course  EY9767.

The update will include the following:

	Replace SPM with DECps
	Remove materials covered in Sys Net I,II,III
	VMS v5.5
	Modularization


Now is the time for instructors to make comments on the existing course
materials.  What is good?  What is not so good?  What should be added?
These comments can be made in the notes file VMS_Performance.  All comments
will be considered.

I will post the project plan in this note as soon as it is ready.

As each new chapter is finished, it will be posted in the following 10 notes
and review comments solicited from instructors.  Please plan to participate in
this process.

Emmalee Tarry
381-1162
SUPER::TARRY

|	Replace SPM with DECps

  Ooops!!  Hold it!  Don't kill it yet!

  DECps is a 1-for-1 replacement for VPA!  That's the module that should
  be replaced, not SPM!  Quoting from Sales Update Vol. 23, No. 6 :

    DECps V1.0 incorporates *ALL* the functionality of VPA V2.1 and therefore
    completely replaces the VPA product. ... All SPM functionality has not
    yet been incorporated in DECps V1.0 so while some SPM customers can
    migrate to DECps V1.0 others may choose to continue running SPM until the
    next release of DECps incorporates all SPM functionality.
    ...
    VPA and DECps cannot run concurrently on the same system.  SPM and DECps
    can run concurrently on the same system.
    ...
    Maintenance and customer support for VPA will continue until March 1993, 
    support for SPM will continue until October 1993.

  There are some performance displays you CANNOT get from anything except SPM,
  so kill the VPA module and leave the SPM module for another year.  (Maybe put
  it in an appendix?)

|	Remove materials covered in Sys Net I,II,III

  NO!  Don't do this!  (Seems that I've said most of this over in the VMS
  Performance conference, but I don't think you've read it yet!)  There are
  many customers that will take this course and *NOT* have taken any formal
  system management courses, much less our new SYSNET string.  DO NOT make it
  a requirement that students must SPECIFICALLY take the Sys Net courses
  before they can take this course.  

  This course, being a specialty course, should provide "Standard Performance
  Management Material", being a *complete* course for the Skilled System
  Managers.  Several of us have suggested that the performance material in
  the Sys Net string should be removed or minimized (too much stuff and not
  enough detail to do any good) in the earlier courses and left in this
  course.

  See note 95.* in this conference for comments regarding pulling some of the
  stuff OUT of SN3.

|	VMS v5.5

  This course hasn't really been *revised* since about V5.2.  There have been 
  a few additions (i.e. vectors and windows), but what the other changes?  For
  example, changes in Lock Management, PRIORITY_OFFSET, size changes for
  IRPs & SRPs, AUTOGEN common-param-file, MMG_CTLFLAGS, AFFINITY_x, Defragger,
  etc.

|	Modularization

  OK, could use a little reorganization, but all-in-all a rather teachable
  course -- let's not rearrange it too much!  :-)

|Now is the time for instructors to make comments on the existing course
|materials.  What is good?  What is not so good?  What should be added?
|These comments can be made in the notes file VMS_Performance.  

  Hmmm.  I'm Confused.  In the VMS_PERFORMANCE conference your note said:

|     This notes file will not be used for the project.  Instead review 
|     comments on the new materials will be solicited in the note file 
|     VMS_CURRICULUM.

  So, do you want 'ideas' over there and 'review' over here?  Seems strange.

Regardless, here are a few ideas...

  The Memory chapter does not 'flow' well.  I've rewritten this module (well,
  actually it is just a cut-and-paste of the existing material) for my lecture
  and it get good reviews on this. 

  I save VPA, SPM and DECps until Friday.  I don't have extended discussions
  of layered products until Friday (other than brief comments) because most of 
  my students don't have these products.

  Chapter one is rather boring and terribly 'artificial'.  Many of the pages I
  actually skip in this module got moved to SysNet3, so maybe we can leave
  those out entirely.

  The I/O module needs a BRIEF intro to the benefits of Striping, Volume Sets,
  Shadow Sets, Defragmentation, etc., not a Sales Pitch or installation
  type of info -- just, 1) What it is, 2) what it does, 3) how it helps with
  performance and 4) where to get more info.  We also need a little info on 
  caching benefits and related products such as the HSC90s, I/O Express, etc.
  I encourage (and actually use) appendices.

 More later...

$  

    Here's my 2p worth...
    
    Re .1 and .2
    
    I completely agree with the points made by Buck in .2 about the
    proposed changes to 9767.  If you take out the material which is now in the
    Sysnet string, it will be a real problem for experienced system managers 
    attending 9767 who have not taken that string (and a real problem for the 
    instructor!). 
    
    (FYI, approximately one third of my students on 9767 in the last year had 
    *not* taken the VMS System Management course (9766E) before attending.
    Also, the UK have not yet implemented the SYSNET string of courses
    although there are plans to look at doing so later in the year.)
    
    Updating the material for all the changes since 5.2 should be seen as a
    priority.
    
    Unlike Buck, I think that most of module 1 is necessary to "set the
    scene" for the rest of the course.  I use it to spell out the fact that
    there are no "magic" system parameters and that application design is a
    BIG factor when looking at performance.  Maybe trim it a bit, but the
    meat of that module should stay.
    
    I'm teaching the course this week, so will post further comments soon.
    
    Geoff
    

Thanks for all the good comments.  There has been a movement to consolidate all
the instructor notes into one file.  I am so glad to get your input, I won't
argue with you about where you put it.  The following notes however should be
reserved for comments on the new chapters as they are posted.

The decision on removing the sys net materials will be up to Andy, but he reads
these notes files also.  

I will check on the VPA DECps and SPM issues.

If you would like to share your cut and paste chapter you can send me a copy
by interoffice mail.   ZKO1-1/D42

Emmalee

    
    
    I would recommend the "tools and investigation" module be either
    significantly re-written or scrapped completely.  At a minimum, we need
    a more realistic example of the "problem" than one program that is
    page-faulting badly, etc.
    
            GJD

  I agree with the rewrite, but I don't think you should dispose of it.
  It is an opportunity to 'tie it all together' that we need to take 
  advantage of!  A small case study, a few 'hot spots' to look at as well
  as "shotgun tuning" when a methodology approach isn't practical.
  
  $

Emmalee,

I have some suggestions:

Chapter 1;

Leave out the review of the standard tools like accounting and monitor etc.
I always do some internals bits in chapter 1, because most of the times
the students do not remember the diff between a job and process and when I
explain some sysgen param. they always ask me where are those parameters
located or what do they change. Include some pictures about the relations
between een JIB,PCB and PHD and I always draw the SHD.

Chapter 2,3,4;

Let them 3 separate chapters.

But they must become "TASK ORIENTED" like the SYSNET courses.
In Holland we have several "PERFORMANCE CASES". We first do the theory
and then we load these cases on the VAX and let the students find out what
is happening on the VMS system.
THEY LOVE THIS APPROACH.

Now adays we only have 1 VAX for this course, so it is not possible to let the
students fix the problems. But we are planning the next stap and that will be
1 VAX for every 2 students. Then we will load the cases on every VAX, the
students can find out what is wrong, and are then able to try to fix the
the situation. This is a nice way of testing the student and see what they
did understand of the lectures.

Chapter 5;

Do we need al the hardware and software explaination of the cluster ?
I suggest to delete this information (it is not a cluster course) and
have only the performance stuff for a cluster in this chapter (monitor,
Distributed lock manager, Ethernet controller information).

Chapter 6,7;

I think it is better to make an appendix out of these chapters, then they
are optional for the trainer.

Chapter 8;

I always split this chapter in four parts.

1) generic part; This one I handle on monday morning together with chapter 1.

2) memory, cpu  and I/O parts; are handle directly after the related modules.
  Why not breaking this module 8 and add the pieces to chapter 2,3 and 4.

Chapter 9;

The student never wants to handle this module It is to much theory.
So again put it in a appendix.


Again:

Today the students wants TASK ORIENTED training with lost of lab sessions.
If the return to there work the must be able to inplement what they have
heard and done in our course.
So design the course this way.


Regards,

Piet Hompus
Nieuwegein
The Netherlands.



Sysnet I,II and III.

I agree with you to leave out the subject which are covered in these courses,
but there is a risk involved. No every trainingcenter will give these courses
the corporate way. In the Netherland we also have change Sysnet II and III a
bit. Especially the part about performance because we think the student are not
ready for it. We try to (task-oriented training) give them hand-on experience
in the available tools like Authorize, install, sysgen, autogen, monitor, and
accounting. Now the go home and they are able to generate reports.

Layered product.

I agree, put the layered products in an appendix like it is done with PCA.
I suggest to mention DECCP only.
Perhaps we must put in SPM,VPA,PCA,DECram and de DEFRAG as well.
Lots of customers do not have these products and do not want us to spend time
on it, so a seperate handout would be nice. I always spend some time on these
products during the week, because the product have a very strong relations with
performance, and 15 min per product is acceptable.


Task oriented training:

As a result of sysnet I,II and III we should make this traing task oriented
too. In Holland we have made some performance cases which we use.
After  explaining the memory module, we start the cases and the student can
investigate the system and try to find out what is happening.
What we do not do is (because we do not always have enough VAX's), we do not
let the students fix the problem. When I do this training and I have enough
VAX's ( 1 VAX  per 2 students) the students go home and really have the
feeling the can investigate and if possible can tune the VAX/VMS system.

We have two votes for improving the Tools and Investigation chapter.  I have
one idea for a problem to add to the chapter. 

I am sure some of you out there have some specific problems that you have used
in your classes.  Please share them with me.

Emmalee

  Good suggestions.  I like the case studies.
  
  But keep the PHD, HIB, etc. stuff on an instructor's page.  I used to
  teach this stuff and found most of my students thought it was
  interesting, but not really useful.  I still cover bits and pieces, but
  the more we have that we can 'demonstrate' or 'explain' with, the more
  flexible this becomes for the instructors.  
  
  Also, I surveyed the last Performance class.  Out of 15 students, none
  had taken SysNet3, 3 had taken SysNet2 and 4 had taken SysManI (old
  course).  Most were 'self-taught' managers, thus we aren't ready to make
  this course the follow-on of SysNet3.
  
  $

  For the Tool and Investigation chapter I 'bug' the system with several
  'bugs'.  When I worked at the Telephone Support Center I wrote down all
  the silly (stupid?) things customers did to their systems and I roll them
  into one lab.  Their objective is to find 5-6 problems.  Actually there
  are several, with a few of them being inter-related, thus finding and 
  fixing one won't solve the problem.  Here is a basic template, modify it
  to fit your CPU and MEMORY needs:
  
$!This is the procedure I use on my workstations which I 'bug' for Thursday's
$!lab in the Perf. Mgt. course.  It makes several assumptions, so make sure you 
$!only use the parts and commands that are useful in your situation.  It *DOES*
$!require that the machine being BUGGED is a standalone machine.  If you use a
$!satellite in a real cluster, you might want to reduce the Disk I/O portions
$!of this code.  I use it on 1 VS3100 in a 15-node cluster with minimal impact
$!on the other cluster machines.
$!
$! BAD_TUNE.COM -- expects P1 to be "SYSGEN" or "DETACH"
$!
$!  If you pass SYSGEN it sets various sysgen parameters and does a 
$!  WRITE CURRENT, then you reboot.  To fix things back merely rename the 
$!  SYS$SYSTEM:VAXVMSSYS.BUG_GOOD to SYS$SYSTEM:VAXVMSSYS.PAR and reboot.
$!
$ set noon
$ write sys$output "P1 =  SYSGEN or DETACH"
$ set ver
$ if p1 .eqs. "DETACH" then goto dtch
$ if p1 .eqs. "SYSGEN" then goto sgn
$ write sys$output "P1 must be either SYSGEN or DETACH"
$ exit
$!
$sgn:
$ mcr sysgen
use current
write sys$system:vaxvmssys.bug_good
Set BALSETCNT 23
Set ACP_HDRCACHE 9
Set IRPCOUNT 500
Set LRPCOUNT 12
Set LRPCOUNTV 12
Set SRPCOUNT 500
Set LONGWAIT 10
Set MPW_WAITLIMIT 10000
Set PFRATL 6
set wsmax 1000
Set WSDEC 173
Set BORROWLIM 11423  
Set quantum 14
write current
$!
$! Borrowlim and Balsetcnt are sized based on my memory and normal job load.
$! Set Borrowlim to about 60-65% of physical memory and set Balsetcnt low/high
$! enough to swap slightly (3-4 processes once students all log in).
$!
$ write sys$output "Make sure PAGEFILES and SWAPFILES are installed and large!"
$ exit
$!
$!
$!
$dtch:
$!
$! This second part is reached by passing DETACH as P1.  It detaches a few jobs
$! and set Highwater/Erase-on-Delete on the SYSTEM and a disk named DISK.  Run
$! this part of the code AFTER running the P1=SYSGEN part and rebooting.  Then
$! let the students log in and find problems using either the decision tree
$! in chapter 4 of the Guide to Performance Management or the module from the
$! lecture guide.  DO NOT let them use VPA or SPM at this time, just Monitor
$! and other 'normal' tools.  Use VPA later to see what you missed!
$!
$ set vol sys$sysdevice /high/eras
$ set vol disk /high/eras
$ run/det sys$system:loginout/time=03:00:00/inp=busy1.com -
   /out=busy1.log/prio=2 /proc="0PC0M"
$ run/det sys$system:loginout/time=03:00:00/inp=busy2.com -
   /out=busy2.log/prio=2 /proc="ERRFMT "
$ run/det sys$system:loginout/time=03:00:00/inp=busy3.com -
   /out=busy3.log/prio=2 /proc="PAYROLL"
$ run/det sys$system:loginout/time=03:00:00/inp=busy3.com -
   /out=busy3.log/prio=5 /proc="RealTimeCollect"
$!
$! Fake names for processes just to make it intereting.  Tell the students
$! that these process represent 'real work' and stopping/suspending them is
$! not legal!
$!
$ exit


   What follows are the 4 BUSYx.COM files. Make changes to these that are
   necessary.  Be creative, but make sure you don't set the priorities too
   high or the students get frustrated trying to get various displays up!

$! Busy1.Com - Lots of pagefaults - Demand Zero & Modified List and COM queue
$ set nover  ! Otherwise log file gets big
$ set noon
$top:
$ show sys/full
$ advise  ! Or you can use ANA/SYS if VPA isn't installed
$ run row_major      ! My copy of ROW_MAJOR exits after 1 loop
$ run column_major   ! My copy of COLUMN_MAJOR exits after 1 loop
$ goto top

$! Busy2.com - Pure SUPERVISOR mode, always in COM (or CUR) state
$ set nover
$ a=1
$top:
$ a=a+1
$ a=a-1
$ goto top

$! Busy3.Com - Lots of disk I/Os, Erase Rate due to Highwater/Erase-on-Delete
$! (I assume a device called DISK and a directory named [SPM25] are available)
$ set noon
$top:
$ back sys$sysdevice:[sys0....]*.com; disk:[spm25]x.x/sav
$ purge disk:[*...]*.*/keep=4/log  ! Hit the ACP_HDRcache
$ purge disk:[spm25...]*.*/keep=2/log  ! Erase rate because of ERASE-on-DELETE
$ goto top

$!Busy4.Com - Some Disk I/O, some CPU, Runs at Priority 5, builds COM queue
$ set noon
$top:
$ cre Busy4.x
qwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnm
qwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnm
qwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnm
qwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnm
qwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnmqwertyuiopasdfghjklzxcvbnm
$ delete Busy4.x;*/erase
$ wait 0:0:01  ! Just so we don't lock up the system
$ goto top
  


Have fun!



$

>
>But they must become "TASK ORIENTED" like the SYSNET courses.
>In Holland we have several "PERFORMANCE CASES". We first do the theory
>and then we load these cases on the VAX and let the students find out what
>is happening on the VMS system.
>THEY LOVE THIS APPROACH.

>too. In Holland we have made some performance cases which we use.
>After  explaining the memory module, we start the cases and the student can
>investigate the system and try to find out what is happening.

Thanks for the input, Piet.  Would you share your case studies with us?
Send them by direct mail to me.

Emmalee

>

  Short update...
  
  I'm currently teaching the Pilot for the new version of the course in the
  D.C. (Landover, MD) training center.  Emmalee is busy taking notes on
  teaching tips, page order issues, extraneous or missing information and 
  I think she's even writing down some of my jokes and stories.  

  This was a 'last-minute' pilot that Emmalee twisted someone's arm to get.
  All the students are regularly scheduled students and I contacted them
  last week to let them know it was a pilot.  So I have 15 'real' students,
  not hand-picked 'reviewers'.  Only 2 have had SysNet 3 and almost all
  have the proper skills for the class.  A few have had internals and some
  had Sys Mgr I & II, so overall a reasonable background. I think Emmalee
  wrote down the number of students that took which prereqs, maybe she'll
  get around to posting them.
  
  Basic thoughts:  Overall the material is well done.  *VERY* few typos or
  editing errors -- I'm quite pleased with this!  The instructor's pages
  are useful, they are NOT merely an echo of the student's page and almost
  every lecture page has a meaningful instructor's page to match! 
  
  All the modules were not ready, but we are covering the new Basics,
  Memory, CPU, I/O, VAXcluster and DECps (replaced VPA) modules this week.
  The Tools, SPM and LAB modules are from the old course.  I'm not sure of
  the Capacity module, but Emmalee is going to teach that part on Friday,
  so I didn't look close.  (Hey, we need some new LAB ideas...please
  submit!) 
  
  Likes and Dislikes:
  
    Glossary - I like it, but not where it is.  After teaching 3 days I
    asked my student's where they wanted the 'definitions', almost 90% said
    at the the back of the entire book.  The others wanted it at the back
    of the chapter.  I tried not to influence it, so I believe we have a
    strong request from the students to move it. 
  
    Index - Wow!  A real Index!  It isn't extensive at this time, but it is
    useful.  I've referenced it myself twice this week and I have received
    a few comments from students -- very nice! 
  
    "Mapping" pages - these are kinda like small "objective" lists
    periodically spread throughout a chapter.  Helps keep a focus, but I
    think it hinders having it in the student materials for a Lecture-Lab
    course.  Move the page to the instructor's side, take it off of the
    student page -- not just this course, but any that use the new format.

    Absence of "Figure 1" and "Table 1" indicators -- This is a departure
    from normal teaching material.  It doesn't seem to make a difference
    since I always referenced diagrams and tables by page number anyways. 
    This is needed for the 'modularity' project.
  
    There is more information in the course, so the material takes a little
    longer to cover than normal.  However, SPM and DECps are appendices, so
    we can cover all, part or none of them depending on time. 

  Prepping:
  
    This course will need to be re-prepped, but it shouldn't take the
    experienced Perf Mgt instructor long -- maybe 1 or 2 days.  A lot
    of the tables and figures have been changed, some rather subtly,
    so I suggest reshooting all your favorite pages for overheads.
    
  Chapter 1 - Basics:
  
    No significant changes.  A few extra pages near the back about DECps
    and such, but a minimal change chapter.

  Chapter 2 - Memory:
    
    Biggest change for the chapters.  Significantly reorganized.  I think
    it flows better, but there are still a few rough spots.  The real
    problem is that each instructor has their favorite tried-and-true
    method of teaching Memory Management.  I was never happy with the last
    rewrite of the Memory chapter.  I thought covering Nonpaged Pool before
    Working Sets and the Free/Modified lists was quite backwards. 
    
    Even with the new material I found myself flipping forward a page or
    two so I could tie together two topics that I normally talked about at
    the same time.  There are a few occurances where a topic is mentioned,
    but for organizational reasons you must say "more on this in a few
    pages" -- For example, you really can't talk about BALSETCNT without
    talking about Swapping...and the Ticker (part of MMG_CTLFLAGS) belongs
    with the Paging topics while Troller (other half of MMG_CTLFLAGS)
    belongs with the Swapper.
    
    It will take a while before I have this module re-mastered, but overall
    I'm much happier with it than the old course.  Warning to the seasoned
    Memory Mgt lecturer: "Practice the new order, but be prepared to use it
    to help you present Memory in your favorite format. It's a tough topic,
    and we will all not agree on it's layout." 
    
  Chapter 3 - CPU:
   
    Changes aren't as dramatic here.  CPU VUP table is an appendix,
    Vectors (for the moment) is the last few pages of the module, Quantum
    was moved a little further back, Mutex discussion is fatter and a few
    new parameters are mentioned (i.e. PRIORITY_OFFSET).  Otherwise
    it is the same old chapter -- easy reprep.
    
  Chapter 4 - I/O:
    
    Several new items (i.e. DECram), reorganized/shuffled layered product
    discussion like Shadowing and Striping, a few of my favorite diagrams
    are missing (don't throw out your old overheads!) and a little more on
    bus/controller contention.  I think we need a discussion of RAID
    technology and then follow it with a discussion of Volume sets, Stripe
    sets, etc.  The stuff on DZ-11s is old, but I like it -- for the
    moment its in there, but...
    
  In brief, a short re-prep will be needed, new overheads, and expect a
  different order of the chapter contents, but the overall structure is the
  same as the last class.  So far, I'm pleased...and many of my students
  are enjoying the class, that's a good sign. 
  
  (More later after the course)

  $

The pilot is over and I think it went quite well -- but then I'm probably
not a good judge at this since this was my first pilot.  Emmalee, how about
some comments?!?
  
  DECps-
   
    This module has replaced the VPA module and it is an appendix.  It  is
    organized based on the part of DECps, i.e. the collector, the advisor,
    etc.  Although the material adequately covered the tool, it didn't
    'flow' for a real discussion.  The data collector part discussed 'work
    groups', so you either have to discuss capacity planning concepts here
    or have covered the capacity module previously.  I suggest taking the
    work group stuff out and adding it as intructor pages or an appendix to
    the Capacity module.
    
    Also, the example of a DECps sample run is an addendum to the appendix. 
    It is rather large as it represents 3 nodes in a cluster.  I think the
    previous course example of selecting 1 node for the report would be
    quite adequate.
    
  SPM -
  
    This is the same module as previous.  No significant change.  At the 
    time of the pilot it was IDENTICAL in content and formatting.  Emmalee
    mendtioned that it may be reformatted, but content probably won't
    change.
  
  Capacity -
  
    Tough chapter, not a topic to end the book with.  I suggest a VERY
    small module, mention the Capacity Seminar and then wrap up with VPA
    and/or SPM.  Example of ABC company's disk space problems is a good 
    starter kit for capacity, but needs more details on the STUDENT page.
  
  Cluster -
  
    Although Locking is a topic without a real home (Is it a CPU, memory or
    IO topic), it fits nicely in the Cluster module.  I usually talk about
    it right about the time I'm covering state transitions.  I explain
    first how locking works on a single node, then how it works on a
    cluster and then tie in LOCKIDTBL, LOCKDIRWT, RESHASHTBL, etc. as well
    as Monitor Lock.  I avoid talking about Lock conversions, as this I
    feel is primarily a programmer issue.  

  Overall, the toughest chapter was Memory Management.  There are too many
  ways to cover it.  I think we need to talk about this one, because of its
  complexity and inter-related topics.  The approach I take works rather
  well and I usually get good comments on it:
  
    The first several pages in the chapter can be taught straight on.  I
    expand the Cache section to discuss 'soft cache', such as FPL, MPL, ACP
    caches, etc. - caches that use main memory.  And I talk about Hardware
    cache - high-speed cache that fits between the CPU and memory. 
    Avoiding details here works best, if you can.
    
    Program locality is a good place to talk about Row_Major and
    Column_Major.  I discuss the 'backwardness' of Fortran vs other
    languages.
    
    I cover the basics of virtual memory, then at 'Regions of Physical
    Memory' I explain that there are "...4 main areas of memory that we,
    the system managers, are interested in...FPL, MPL, Working Set and
    'everything else'".  I take them quickly to the Show Memory and explain
    the 'everything else' is in the last line which says something like: 
     
    "Of the physical pages in use, 8542 pages are permanently allocated to
    VMS" We'll talk about it later.  At this point I start working on the
    FPL, MPL and Working Sets in a 'story-line' which consists of processes
    getting pages from the FPL, adding to working sets, growing working
    sets and minimal page-removal, up to, but not including AWSA-decrementing.
    (I think the FPL and MPL pages should be further to the front, probably
    near Paging and Page faults, and the diagram of such (currently on page 
    2-37&38) should be in front of ASWA decrementing.  MMG_CLTPAGES feature
    called TROLLER should be AFTER Swapping is discussed and the the Ticker
    should be immediately before or after AWSA decrementing.
    
    I also found that having the monitor displays at the back of the
    section a bit inconvenient.  To talk about Hard-faults I like to have
    them run MONITOR PAGE and show the info.  
    
    The diagrams on pages 2-61 and 2-19 are funtionally equivalent.  If
    were are going to keep them, they should have similar formats and the 
    the first one should NOT show page sharing since 2-61 is where this is
    really discussed.
    
    DECwindows should be the LAST topic of this module, not shared images.
    
  As you can tell, there are lots of isues to still be worked out here, we
  need to get a few other training centers to get copies of the Memory
  module and 'test it out' during your courses and get input back here.
  Emmalee, can we get an on-line copy of the Memory Module I taught for the
  others to 'test' and review?
  
  Quick comment needed...if the MONITOR or SHOW display has not changed
  since the version it was created on (carefully examine each), then would
  merely updating the DATE be sufficient, or do we need to generate new
  displays with new numbers??  I liked the old displays, new dates (1991 or
  1992) was all I really wanted for many of these.
  
  $
  
    

As most readers of this conference know, the version 5.5 performance course was
piloted in Landover the week of June 15.  I was most encouraged by the pilot.
It was most beneficial to participate in a pilot was that conducted as all 
pilots should be conducted.  That is:

	Taught by an experienced instructor

	Attended by typical students

	Observed by the course developer

	Materials in student ready condition
	
	Time to revised the materials after the pilot.

I hope that in the future that each and every new course or course revision
will be piloted in like manner.

I really appreciate the time and effort Buck has put into reviewing and
prepping the new materials as I appreciate all the comments and suggestions
that have been made in the notes files. 

I have not had the time to respond to each and every comment in the notes file,
but be assured that every one has received attention and that most have been
implemented.  After I get the course to the editors, I will take the time to
respond to each reply.  In the mean time there is still some time ( not much )
to get in your comments on the chapters as they are posted.

    I did make some notes at the pilot regarding previous classes taken by
    the students.
    
    	Sys net III     2
    	Sys Man II      4
    	Sys Net I & II  7
    
    A point of interest.  One student who has now taken Sys Net II and III
    and Performance, insists he is going to take Sys net I because he wants
    to finish the entire string.  What this says to me is that the idea of
    naming the courses I, II and III   IS WORKING!
    
Some comments.  

I moved the TIME factors to the instructors page.

The chapter terms remain in the front of the chapter, but I think they will be
moved to back of the chapter.  Is this really such a big issue?

The course now has 7 chapters with 2 appendices.  I am still hoping to have
some time to work on the labs.
    

|I have not had the time to respond to each and every comment in the notes file,
|but be assured that every one has received attention and that most have been
|implemented. 

YES!  I need to mention that I was most impressed.  Emmalee had extracted all
the notes we had posted, and had marked up the listing indicating whether it
was implemented, fixed, open for discussion, not implemented, needs research,
etc.  I wss quite pleased to know that the reviews I've been poking in here
(as well as the other people's comments) are being seriously considered.  I 
think this "Notes Thing" might actually be working -- but it sure takes a lot
of typing. :-)

$

    Some comments on the new LL format.
    
    The objectives listed in the front of each chapter map to the sections. 
    The wording of the section title should be close to if not exactly the
    objective.
    
    Each section begins with an introductory page that contains only one
    block, the Overview.
    
    The Overview tells why the target audience needs to know the material
    in the section and then lists the subsections or map titles in the
    section.
    
    The next student page begins the first map.
    
    The idea is for the instructor to use the Overview to introduce each
    section.
    
    Now having said this, I must admit the Overviews in the course are very
    thin.  I wish I had the time to improve them.  I am sure the format
    will undergo some evolution with instructor suggestions.

>As most readers of this conference know, the version 5.5 performance course was
>piloted in Landover the week of June 15.  I was most encouraged by the pilot.
>It was most beneficial to participate in a pilot was that conducted as all 
>pilots should be conducted.  That is:
>
>	Taught by an experienced instructor
>
>	Attended by typical students
>
>	Observed by the course developer
>
>	Materials in student ready condition
>	
>	Time to revised the materials after the pilot.
>
>I hope that in the future that each and every new course or course revision
>will be piloted in like manner.

I couldn't agree more!!!!!!!!!!!!!

Last Call!!!!

I have posted the almost final revision of the memory management chapter in
ES$REVIEW.  Please take a look at it and get back to me tomorrow with any
changes.

Ignore typos please.

This is chapter that is the most changed from the existing course.


Emmalee

This is 3-July and I have just sent chapter 1-7 and the two appendices to the
editor for FINAL work.

I have posted the latest runs of chapters 2-5 in:

	SUPER::$1$DUA6:[ES$REVIEW.VMS_PERFORMANCE_V55]

Some instructors might like to pull them and begin to prep.  The course will be
available by the end of July, but I am not sure how soon it actually gets out
to the field.

I want to thank Buck Trayser for his considerable efforts.  Buck reviewed every
chapter several times and taught the pilot. I also appreciate the comments
others made in the notes files. 

I was not able to do everything I wanted to do for this update due to the lack 
of time. I am planning to work on at least two sections after the course is on 
the shelf and post the revised materials in the notes file.  Instructors will 
be able to pull and present these materials if they wish.

After I come back from vacation, I will have 1 week in which to work on the
post test and the laboratory exercises.  If you have any lab ideas send them
in now.

I am going to petition to have the old notes file VMS_Performance opened  for
comments on the new course.  Also I would like to build a distribution list of
instructors who teach the performance so that I can send updates and other
interesting stuff that comes my way directly.

There are some exciting things coming in performance so stay tuned.

    
    
    I'm going to be "on the road" for the next couple of weeks, and then
    come back to teach Performance (last week in July).  Will that be the
    "old" course or "this course"?
    
        GJD

  I haven't seen an "on the shelf" date yet, but I suspect it won't be
  ready until August or later.  HOWEVER, as soon as the course is finalized
  an Instructor's Guide should be available on-line, I'd suggest getting it
  and prepping for it.  
  
  There is nothing that an existing Performance Instructor should have a
  real problem with except for the new DECps module and the rewrite/reorg
  of the memory module.
  
  $

    
    
    Danke!!!!!
    
       GJD

    The performance management course has been updated to OpenVMS v.5.5

    The new course name and number is:

        OpenVMS System Performance Management  EY-L565E



    I have placed a copy of the instructor guide in the ES$REVIEW directory.
 
	SUPER::$1$DUA6:[ES$REVIEW.VMS_PERFORMANCE_V55]EY-L565E-IG-0001.PS



    Time for this project was severely limited.  Not all promises were kept.
    One promise that was kept was that every instructor comment that was put
    into the notes file was considered. I hope that you will continue to
    make comments in the HARDY::VMS_PERFORMANCE notes file.

    Major changes:

	The chapter on SPM and a new chapter on DECps were placed in the
	appendix.

	This means the course now has 8 chapters rather than 10.

	The new LL format was used.

        The material on terminal performance was placed in a chapter appendix.
    

   I would like to have a mailing list of all the instructors worldwide who are
   interested in this course.  This list will be used to send announcements
   directly by mail regarding this course.  Please send a mail message to:

	SUPER::TARRY



   One of things that did not get done due to the time constraints was to make
   any changes in the laboratory material.  I hope to post some notes in this 
   notes file regarding laboratory material.  If you have any pet labs please
   share them with everybody.
    
    There was a request to return to using the notes file 
    HARDY::VMS_PERFORMANCE.  This notes file has been reopened and notes
    established for each chapter for comments.