| Title: | POLYCENTER Console Manager |
| Notice: | Kits, Scans, Docs on CSC32:: as PCM$KITS:,PCM$DOCS:, PCM$SCANS: |
| Moderator: | CSC32::BUTTERWORTH |
| Created: | Thu Aug 06 1992 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 1541 |
| Total number of notes: | 6564 |
I have a customer who has an extensive VCS installation (several VCS
stations) monitoring a very large, complex, multi-cluster environment.
My estimate is that currently about 99% of the occurrences of red icons
as well as error windows automatically popping up are nuisance
messages, i.e. the operator takes no action other than to click away
the message.
My question is, what are some other people's experiences in this area?
What is a reasonable expectation as far as "accuracy" of severe error
flagging at other customer sites? 10%? 30%? 50%? It has to be more
than 1%!
My customer's operations staff end up often not paying attention to
and/or totally missing the really important messages because they spend
so much of their time clicking away the nuisance red icons! Also,
they'll sit there for hours on end ignoring a screen full of red icons
because they are 99% sure they are not important!
Also, what are some of the easiest, most common, and/or most productive
ways to refine the filtering? We are embarking on a project to clean
up error reporting at the applications end as well as correct some
operational process issues. However, some of the 3rd party
applications can't be changed and I'm betting they will still have a
lot of nuisance messages.
I'm a project manager with a technical background. I'm a quick study
but haven't customized VCS myself in the past.
Also, are there any features of Console Manager that make it inherently
easier to get a higher percentage of legitimate red icons?
Thanks! Mike
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 547.1 | easily modified | ZENDIA::DBIGELOW | Innovate, Integrate, Evaporate | Tue Jan 10 1995 03:34 | 17 |
Each event can be customized depending upon the severity of the event.
If the customer doesn't care about certain events, they can be
'reconfigured' out of the scan profile. Alternatively, the priority
can be lowered so that the icon's turn green. This is probabily easier
to do in PCM than with VCS.The absolute easiest way is to use the
editor to customize each event. This will probabily take you the
better part of a morning. Alternately, you can export the data base
to an ascii file, edit it directly and then import it back. This
requires a little understanding of how the database works, which is
not teribbly difficult, in order to implement.
If your customer is tired of VCS crying wolf, then I suggest you
modify the events so that red really means take action. Just one point,
whne these events were first classified, the critical events were
indeed deemed to be critical.
Dave
| |||||
| 547.2 | Any data on other customers? | LOCH::KNAUER | Tue Jan 10 1995 22:19 | 9 | |
Thanks for the response and the ideas. Any other people out there have
any numbers on what a reasonable expectation is for % of "legitimate"
vs. nuisance red icons occurrences? How long and hard (weeks? months?
ongoing?) is the refinement process to get rid of all the nuisance
messages? Do your customers ever really bother to follow through and
do that? How much is the usefulness diminished by all the "crying
wolf" messages?
Mike
| |||||
| 547.3 | OPG::PHILIP | And through the square window... | Wed Jan 11 1995 09:49 | 37 | |
Mike,
The number of critical events will depend upon what each particular
customer requires, so any numbers I could give here would really be
irrelevent. Now, having said that a %age value for legitimate vs
nuisance would be that 100% of all critical events should be just
that, critical, and that in an ideal situation, these should be very
rare occurences (My gut feeling would be that if I got more than say 1
critical event a week, I would be worried)
Refining the notification of these events is a never ending process,
new situations which require notification occur all the time, especially
if a system changes its application mix, in effect, each time an event
occurs that the customer is interested in, then that event should be
included in the PCM scan set, any which the customer is notified of and
are just "nuisance" should be immediately removed from the scan profile
so that they do not occur again.
Some controls your customer should put in place in order to ensure that
the events being notifies are the correct ones...
1) Any application acceptance criteria the operations department has
should include the fact that the application development team provides
a scan profile for that application (after all, they are the ones who
generate the messages).
2) It is the responsibility of each member of the operations staff to
highlight events which have happened, but for which there is no
automatic notification, this notification should then be set up. So,
there should be a well defined process which is easy to use that
enables new events to be defined.
I hope this helps.
Cheers,
Phil
| |||||
| 547.4 | Putting burden on Application Group: Good Idea! | LOCH::KNAUER | Wed Jan 11 1995 17:46 | 13 | |
Thanks for the response, ESPECIALLY the idea of Operations pushing back
on the Applications group to provide the scan profile for each
application as part of the acceptance criteria -- I don't know why I
didn't think of that myself!
Also, you're right that achieving and maintaining a high percentage of
legitimate red icons (vs. nuisance ones) will be ongoing and require
constant vigilence, since this customer averages at least one
configuration change and one application change/re-release per week!
Thanks again,
Mike
| |||||