Text 57, 70 rader
Skriven 2005-04-24 11:29:32 av Mike Luther (1:117/3001.0)
Ärende: Fool on Site level?
===========================
In my humble opinion we are so blessed with the stability of OS/2 and the
PEERLAN capabilities in it that for small networks we don't think about what a
couple of my friends call the need for a 'Fool-on-Site'. That's humor of course
as to the choice of words to describe a really knowledgeable networking person.

If set up decently, and the network traffic isn't really high, such as for
really minimal text-based data access to a good few shared file's on a shared
OS/2 peer box, my experience bears that OS/2 runs a VERY long time without much
real networking service or troubleshooting.

I've noted that in, say six or so workstations tied in IBMPEER to one 'peer
server', actual complete failure mode for server access to shared files is
actually on the order of, at one site, two hours of total failure in now seven
years of such service.  That for all the hardware,software, networking (lack
of!) service and the like,

I have very similar service history on my remote communications site as well. 
And at my master site, plus all my personal efforts, that's also a function of
full linear pig-iron power supply relay rack cases, APC's full relay rack smart
UPS boxes, Rose switch professional KVM equipment and so on.  So proper choice
of a right, there is no THE right,motherboard, case power supply and hard disk
vendor sources is sure part of the mix.  And yes, Virginia, even the hubs,
modems, and everything are behind APC protection too.  Plus with surge
protection everywhere,even in non-remote site non really ligthning prone
service, thank you.

And no, that magnificent down time minimum doesn't count planned access for
service work on things.  It is only the actual catestrophic failure mode time,
where all the boxes have APC's power protection and so on.

BUT ..

At some point, where the total number of nodes in a network grows and grows,
what is your feeling that a decently trained, in humor for the name here,
Fool-on-Site that really has network hardware and setup skills, plus hardware
tracing toolsets for CAT-5 and so on, is needed?

Sure .. that's a function of how mission critical is the reason for the whole
system!  And that's also a serious function of how heavy the data load is for
what is crossing the LAN.

But, at what point, in, say a 'doctor office', with what level of connectivity
to outside world as to data type and terminal use, must we really contemplate,
say WSEB as opposed to IBMPEER?  And at what point,from an opinion standpoint,
do you feel that there has to be someone locally there on site to really debug
broken NIC's, clean case fans on a regular basis because nobody there has the
sense to even take a look at stuff like that and even though you TELL them to
do a little PM, CHKDSK and whatall, they never do?

When should one begin to service connected boxes with a switch, instead of a
hub, in your opinion?

Let alone ever write down any form of written log on every box in the system so
that when the Fool-in-Travel, or Fool-on-phone calls, he, she or *it can have a
prayer chance of salvation for a catestrophic system such induced failure?

Opinions, based on experience, if anyone will share grief, pride, or experience
from proper training?  I'm contemplating s couple of 100 plus box operations at
some point in the future.  I'd like to hear of experience recaps specifically
for OS/2 systems.

Thanks..


--> Sleep well; OS/2's still awake! ;)

Mike @ 1:117/3001

--- Maximus/2 3.01
 * Origin: Ziplog Public Port (1:117/3001)