Re: Informix/Oracle comparison on Data Storage terminology

From: Jonathan Leffler <jleffler_at_earthlink.net>
Date: Wed, 04 Dec 2002 07:01:41 GMT
Message-ID: <3DEDA7E3.2050009_at_earthlink.net>


Malcolm Weallans wrote:
> Jonathan,
> I don't think people are saying that onstat is a replacement for
> onmonitor. Some people aren't as fortunate as us and are able to
> use onmonitor. They might be running WGS or even NT versions of the
> product. To them onstat is "the tool".

OK. I wasn't suggesting that onstat is the replacement for onmonitor; merely that Neil's remark about "does anyone with more than one month's experience actually use onmonitor" makes even less sense as a question if you replace the word 'onmonitor' with 'onstat'.

> I don't think that onmonitor is perfect. Our experience of how to
> manage Informix systems has grown since the days of Turbo, but the
> onmonitor displays have not. I would like to see Informix commit to
> extending onmonitor by putting in some really useful status
> displays such as the table access profiles, user usage profiles,
> and accurate info on log log usage.

Unlikely to happen. It is perpetually slated for removal, periodically removed from beta versions, and reinstated in time for GA. However, ON-Monitor is fundamentally not a happy program; it is an I4GL application that happens to connect direct to the system shared memory -- ick!

> We could even write in in 4Js so that is was then available as
> character - very close to the machine, GUI over a windows network,
> and web - and guess what -- we wouldn't even need to change the
> source. But maybe that idea is too novel:-)

The source code would have to be changed drastically. It is an odd mixture of I4GL and plain old C code. The I4GL bit could be recompiled easily enough, but the C code is not about to be released; far too much proprietary data about the structures in the shared memory etc. If anything like this was to occur, ON-Monitor would have to be rewritten to use the other ON-* utilities to do the real work, and just leave ON-Monitor to interpret the output. Of course, reading files and handling strings for parsing the output from onstat, oncheck, onspaces, etc is not I4GL's strongest point.

I don't think that this is likely to occur, unfortunately.

> -----Original Message-----
> From: Jonathan Leffler [mailto:jleffler_at_earthlink.net]
> Sent: 03 December 2002 04:46
> To: informix-list_at_iiug.org
> Subject: Re: Informix/Oracle comparison on Data Storage terminology
>
>
> Neil Truby wrote:
>

>>Does anyone with more than one month's experience actually use

>
> onmonitor?
>
> Yes. I typically use it once - when I initialize an instance.
> Amongst other things, I can't be bothered to remember what the NETTYPE
> entry is supposed to look like (there isn't a prototype entry in
> onconfig.std), so I use ON-Monitor to ensure I get that bit right.
>
> After that, I use it seldom, but on occasion I do - database owner or
> logging status can be useful on occasion, for example.
>
> [And I've seen some suggestion that perhaps 'onstat' was intended
> instead of ON-Monitor--a question which makes much less sense to me; I
> don't see how anyone who has anything to do with IDS administration
> can avoid using onstat on occasion (and typically much more frequently
> than just occasionally if performance is ever an issue). It's also
> one of the faster ways to find out if the server is still up.]
>
-- 
Jonathan Leffler                   #include <disclaimer.h>
Email: jleffler_at_earthlink.net, jleffler_at_us.ibm.com
Guardian of DBD::Informix 1.04.PC1 -- http://dbi.perl.org/
Received on Wed Dec 04 2002 - 08:01:41 CET

Original text of this message