🚀 go-pugleaf

Forwarded/Reply created @ 16-NOV-1992 13:03:38.20
============================== Original Message ==============================
>Date: Mon, 16 Nov 1992 12:06:48 -0500
>From: Colin Strutt LKG2-2/T2 DTN:226-5340 <strutt@took.enet.dec.com>
>Subject: RE: databases in management products
>
>Bob Boyd writes:
>
>> Right now it looks to me that the only way we can export the data to SAS is
>> by running MCC commands to SHOW the data and then reformatting it into some
>> form we would then use for SAS processing.  Is there anyone else on the
>> list who has already conquered this one?
>
>I would be interested to know exactly how the data should appear, if it is
>not in a SQL form.  For example, would you like it to be encoded in the same
>way that it is encoded in the MIR? (Seems unlikely, as the program(s) that
>would decode that information would have to be quite complex).  Would you
>like to see it in ascii - and if so, formatted how?  (This seems similar to
>what you would get with a 
>	SHOW xxx, TO FILE yyy
>except that your ascii parsing would be a little different).  Is there
>another approach that I've omitted?
>
Colin,

I'm a little bit at a disadvantage here since I have not gotten real
interested in looking at the guts of records in the MIR.  So, I don't 
really have a clear sense of how gnarly it might be to decode the stuff
that's out there.  I can sort of imagine from looking at the DECmcc System
Reference Manual that it might be interesting.  I assume the complexity 
you're suggesting is due to the degree to which the information is encoded
-- not because of the organization of the database.

I'm curious about what the schema for the data looks like with Rdb or
Ingres.  I've looked at the Historical Data Services Use manual and see
that the records are fairly simply organized, but could get quite large --
especially with SNMP or CMIP managed objects.

This does sound like an interesting problem -- I can think of some
alternatives but all of them have drawbacks of some sort.  Seems like one
major complication comes around the possibility of having data from
disimilar entity types being exported to the same data file(s).  Because of
this, it seems that all of the identifying information has to appear in the
record itself instead of appearing in some kind of header block.

Of course if the file(s) could be separated on the same basis as the tables
-- then it might be feasible to have a header record at the beginning of
the file that gives the structure of the records in the rest of the file.
Then the rest of the records would be an ASCII representation of the data
with some sort of delimiter between fields -- essentially just another form
of a table.

This might be a good compromise between the ASCII format of the SHOW 
command and a pure "internal" format.  Comments?

>
========================== End of Original Message ==========================

Bob Boyd	Unisys/EPA, RTP NC	919-541-4441
rbn@c3po.rtpnc.epa.gov