Conference azur::mcc

    
    
    It sounds like one thing that we could do is to include Alarm
    templates for all of these rules with ELM.  That way other customers
    would not have to go through the same setup pains that this one did.
    It's probably too late to add this in the current release, but it
    should be added in the next one.   
    
    One issue with this is funding.  There is currently no funding in the 
    Networks Infrastructure org. (used to be LAN BU) for ELM development.  I am
    currently preparing a maintence release of ELM to make it compatible
    with the MCC V1.3 release (with a lot of help from some MCC engineers
    who have graciously donated their time) but after that things are 
    uncertain.
    
    As far as Alarms performance someone in the MCC group will have to
    address this.   I don't know enough about how Alarms works to address
    the performance issues.
    
    I am passing your note onto Mike Bouchard the ELM Product Manager
    so he can see first hand how customers are struggling with the
    transition from ELMS to DECmcc ELM.  
    
    Regards,
    
    Bob Harokopus
    ELM Project Leader
     

Hello,

A LANman I'm not but there may be better ways of setting up the customers
domains and alarms for reducing the number to below 480. I would tend to think
the user really doesn't need to monitor all the attributes described in .0 but
I can't help with any specifics. 

With every version of MCC my first priority is to understand what wildcarding
enhancements have been made to ALARMS. Your user using 480 alarms might not
be using widcards to their best advantage, and also may me in a maintenance
nightmare for upkeep. This will probably wear on him over time. After 
understanding fully the wildcarding capabilities of EACH TYPE of alarm 
decisions can be made on domain structure and population. All OUR DOMAINS ARE
BASED ON ALARMS FUNCTIONALITY/WILDCARDING. This is clearly a tradeoff for what
we'd like to 'view' in the maps but has made our lives much richer.

Example:

We monitor all NODE4s and their CIRCUITs for availability. CHANGE_OF rules 
would be the first choice for a rule but the wildcarding isn't as strong as
the standard rule. Using CHANGE_OF we would require at least 100-125 rules to
get the job done. TRADEOFF time. By populating the domains carefully to ensure
each router only appears in one domain (ensures no double polling) we have 
set up standard wildcarded rules with:

expression=(node4 * circuit * substate<>none, at every 0:15:0)

We have twelve of these polling domains so their are only 12 rules to do all
NODE4 and CIRCUIT availability polling. The TRADEOFF is that the depiction of
other routers (which we don't want to poll) in the maps aren't put in as NODE4s
but as reference entities. The other drawback of having constant mail during
the outage is resolved within the procedure firing. This solution provides us
with a zero maintenance alarm system (other than adding/deleting from the maps)
with few rules and therefore less overhead.

So domain structure and the maximized use of wildcards are key to reducing
overhead. To be honest I'm impressed your customer has got as much running as
he does with a 3100/32m system. The user appears to understand the tool well
so I hope this info isn't moot.

good luck,
brad...

This is  an issue I raised some time ago.  Why are we forcing the customer (read
internal OR external) to give up something THAT WORKS FINE and move to an 
extremely complicated environment, namely DECMCC??  

DECElms works, has worked and could be mad to continue to work just fine with 
the FDDI products.  This person is lucky that he has the time, and talent
to weed thru all of the documentation to make MCC work.

Why not just make DECElms work with the new bridges??

Ed

    Greetings,

    I was looking for a place to put this and it seems that this is as
    good a place as any....


    I am the CSC account rep for Oracle Corp. and have been asked to pass on
    the following message by their DECnet network manager.  He is
    currently in the process of installing some DECbridge 6xx products
    along with other various FDDI equipment.  He is VERY dismayed that
    DECmcc is the only supported management tool for these products.

    The customer stated:

    DECmcc is a very powerful product and has many bells & whistles ...
    but ... its too hard to use (he thinks its has an extremely large
    learning curve) and it limits their real-life network management
    environment.  He is concerned about the following:

    1) High learning curve...part time network managers don't have the
    time or resources to learn DECmcc.  Even experienced TCP/IP
    managers freeze up when trying to use DECmcc.

    2) You need a dedicated Workstation which equals high cost and
    limits the number of network managers that can do work at one time. 
    He says that Oracle has 7 network managers and due to things like VMS
    licensing issues not all seven can be on the DECmcc machine at once
    and due to costs constraints having 7 DECmcc Workstations chained
    together (if this were possible) won't work either.  Also the
    Workstation environment limits the work that he can do from home
    via dial-in modems.

    3) DECmcc is too complex and thus too slow for troubleshooting
    critical network problems.  With DECelms it was very easy and
    strait forward to troubleshoot problems.  He would like to see a
    DECelms environment available to complement DECmcc.

    4) The customer feels that due to the cost and complexity of a
    DECmcc environment, and the fact that a less complex and user
    friendly environment (like DECelms) is not going to be available,
    the investment in DEC network products (bridges, Concentrators etc)
    that Oracle has made will be "money down the drain". He thinks that
    DEC network hardware is top notch but DEC's network management tools
    make it too hard to use that great hardware.

    The customer expressed to me that their is a large contingency of
    TCP/IP fans at his site and they are heavily recommending that DEC be
    eliminated as a network vendor due to the cost/complexity of a DEC
    network management environment.  He is fighting for DEC but he
    thinks the battle is being lost!

    He would like to know if there is a chance that he can get a version
    of DECelms (even a patched one) that will let him manage his
    DECBridge 6xx products.

    If their are any replies or if any one wants to talk with this
    customer please feel free to post them or call me.

    Thanks,
    Stephen Rochford
    CSC/CS DTN 592-4546

    I will pass all of your comments onto product management.  This has
    been an issue for quite a while now.  However, I doubt ELMS will ever
    be extended to manage the DECbridge 5xx and 6xx.
      
    When the DECbridge 5xx and 6xx series was in development a decision
    was made to use DECmcc as the management platform since that is the
    company's strategy around enterprise management.   We simply did not
    have the resources to do both ELMS and MCC.   In addition,  MCC's promise
    is to eliminate the need for all of these point products and instead
    offer an integrated package for network management.   
                                 
    I think what needs to be addressed now is how to make MCC and ELM meet your
    customer's needs.    If our customer's are indeed afraid of MCC then
    this is a bigger issue than just bridge management.
    
    -Bob  
    

Thanks to everyone for the technical suggestions, sympathy and 
respect for my customer. Yes, I too admired the fortitude and
talent to recreate 480 alarms. 

I'm going to do two things:

	1) get information to the customer on DECmcc V1.3. I'm
	   thinking the "TCP/IP reachability" polling feature
	   may help along with toning down the number of
	   alarms

	2) Pull this thread of conversation out and submit it to
	   the "Network Management SIG" of the Network Partners
	   program as an issue to raise to the PBU.

Thanks again, Dave

    As a side note, thanks for posting the list of attributes you
    were monitoring.  I will look into extending the alarm manager
    knowledge base that we supply to include alot (if not all) of
    these so that the creation of the rules should be drastically
    simplified.
    
    Also note that we are working on an interface into the alarm
    manager (using PM to PM communication which is not generally
    available yet) such that you can select an icon on the map
    and request from the Operations menu to "assign template" (verb
    name DEFINITELY subject to change). At which point they will
    be given a list of templates that apply to the entity class
    selected, and allowed to apply the template to the entity to
    quickly create a rule.
    
    Will this help with some of the learning problems? (FYI, templates
    are self documenting providing the if's when's and why's of using
    the template right in them)
    
    

Hello,

We have a problem here which has never been addressed....alarm scalability.
MCC starts polling for INDIVIDUAL attributes for EACH alarm. When monitoring
large environments where the customer is looking to keep track of multiple
statistics/errors/whatever the alarm rule required are individualized and sent
off to poll everything seperately per rule. While this is very generic and 
powerful it doesn't scale as illustrated in .0. MCC should somehow design
'something' which allows a single poll to be used for multiple ALARMS and/or
historical data collection. Make the polled data available for multiple uses
rather than stripping it down to a single attribute for a single rule. Tough
job, but needed if you want to reduce MCC resource consumption over the long
run.

best regards,
brad...

RE: .7

>    Also note that we are working on an interface into the alarm
>    manager (using PM to PM communication which is not generally
>    available yet) such that you can select an icon on the map
>    and request from the Operations menu to "assign template" (verb
>    name DEFINITELY subject to change). At which point they will
>    be given a list of templates that apply to the entity class
>    selected, and allowed to apply the template to the entity to
>    quickly create a rule.

  This sounds like the right track to me .. and its called RELATIONSHIPS.

  You select and Entity, press MB3 to get the pop menu, and see RULES
  on it with an arrow.  Move to the arrow and see all the things you
  can do with rules; like, create, enable, show ...

  The Create menu item will bring you to the Alarms Manager where you can
  create Alarm rules via the template mechanism.  Using the Show menu item
  you can SEE RULES CREATED FOR THIS ENTITY with no fuss or bother.

  Jill .. sounds like a winner to me .. next release of the Alarms Manager ?

  /keith
 

Jill:

  What I would like to see is more along the lines that Keith suggested, but
  slightly different.

  The user selects an entity on the map, presses MB3 and selects "Alarm Rules".
  Then the map window would display a list of the alarm rules that are RELATED
  (we can discuss the exact meaning of "related" in another forum) to the entity
  selected (and only those rules -- *not* all the rules in that domain).

  Then, any of these rules could be selected and operated on.  The operation
  menu would, of course, change to display the operations valid for alarm rules
  (i.e., create, delete, show, set, create template, etc.).  This is exactly the
  same paradigm used to manipulate entities on the map!  

  This is what some of us have been calling RELATIONSHIPS.  It is analogous to
  products like Hypercard/Hypertext.  Capabilities like this exist in the 
  Bookreader product.  In more technical terms, we need for the IMPM to allow 
  users to move (i.e. navigate) between entities via relationships other than 
  the containment relationship (i.e. parent -- child ).

  You've all heard me say this before -- and you'll probably hear me say it 
  again!

  Would anyone else out there like to see this kind of capability?

                                                                       - Jerry 

    Re: Alarms Relationships -
    This is a must for me.  I have been asking for a means of managing
    hundreds of alarm rules for some time.  It would be great to also be
    able to include the category on the map, double-click on it to expand
    and see all alarms for that category.
    
    Alarm relationships is a great idea, something other vendor products
    have, and it is something my customers "ding" me on rather often.
    
    
    RE: Bridge management using DECmcc:
    If you want to limit the sale of bridges, allow them to only be managed
    via an
    expensive-management-product-that-requires-lots-of-resources-and-works-
    on-one-vendor's-products.
    
    
    -Dan

>    Alarm relationships is a great idea, something other vendor products
>    have, and it is something my customers "ding" me on rather often.


        *Which* vendors have this ?  Vendor & product names + brief
description would be helpful.

/doug

Doug:

  You bring up a good question.  I am also curious to know if any other 
  vendor has this capability.

  I personally have not seen any other Management product with the power 
  of DECmcc's generic, user-definable alarm rule based scheme, and I also have
  not seen any other product that does what is being proposed in the Relationship
  idea.  Remember the Relationship idea is not limited to just relating alarm
  rules and notifications to their corresponding entities!   It could make the
  entire IMPM easier to use while at the same time providing more management
  power to our customers.  Relationships will do for DECmcc what context 
  sensitive HELP did for HELP; what HOT SPOTS does for Bookreader; and what
  HyperText does for Library research!  I am convinced that we need this in 
  DECmcc.

  Four years ago there was very little managment software that provided 
  user-definable rule (or expression) based management functions.  Today I'm 
  seeing more and more of it.  For example, Remedy is touting their
  user-definable expression capability in their graphing functionality.
  The competition is catching up in this area.

  Are we going to wait until some other vendor does something like what we 
  are proposing for the Relationships idea and let them take all the glory --
  not to mention the profits!!   

  Now I know that this is a radical idea in DEC these days, but why can't we 
  do this first!!

                                                                        - Jerry
    
  P.S. Another thought to ponder:

       Microsoft is doing in Cairo something very similar to what we have been
       doing in DECmcc all along.  Why are they so successful with these ideas
       while DEC struggles in the market place??????

    NetLabs has relationships.
    
    
    Also, Until Digital learns to take a unique approach to marketing and
    advertising, we will not be a contender in the market.  Perceptions are
    everything.  DECmcc/MSU/DECmcc/EMA/mcc/MCC/POLYCENTER Network Manager
    200/400/POLYCENTER SNMP Manager 300/POLYCENTER
    Framework/DECmcc-SMS/DECmcc-EMS/DECmcc-BMS/DECmcc Director/...  
    
    Just look at this blob of Alphabet soup.  Where is the customer's
    solution?  
    
    Digital is #1 at stealth marketing.  We're so good that our sales reps
    can't even figure it out.
    
    The problem is not just with DECmcc.  We have changed the names of so
    many products it is amazing.
    
    No matter how hard you guys work in engineering, and no matter how much
    better our products might be than the competition, the bottom line is
    perception.
    
    We need clear and concise messages.
    
    We need to advertise until it hurts.
    
    We need to provide customer solutions and tell everyone about it.
    
    A low-cost PC platform (mcc-lite) with limited functionality would give
    us name recognition and a foot in the door to sell the big stuff.
    
    dh