Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: (ASCEND) Does anyone know how to do Port Management - ie: SNMP ...
On Wed, Mar 11, 1998 at 04:43:20PM +0000, Adam Neat wrote:
>
> Ive been trying to find a solution for this for ages - but it seems I'm the
> only one who wants to know about it.
You're probably digging into the wrong direction.
> We want to be able to look at a client's port and see their utilisation and
> total byte count.
>
> We also use Cisco's and we can do this from our management stations.
Ciscos have fixed interface<->peer mappings which have other shortcomings.
The real problem is neither Cisco nor Ascend but the fact that NMSs are
too unflexible to allow you to ask this in any other way than by fixed
interface. This doesn't make the concept better, it is just a "status quo".
> I know there is some bug / error with the Ascend SNMP implementatin where the
> indexes are never the same, but, there _HAS_ to be some other method of saying
This is no bug. It's completely left to the implementor of a system to
whether he implements a static interface<->peer mapping (let's call it
the interface based approach) or a dynamic one (call this the profile
based approach). I _like_ the profile based approach. Cisco is coming
from the backbone routing world and has boxes with a limited, quite fixed
peer mapping and thus has always used and continues to use the interface
based approach. Ascend, however, is doing primarily NAS business and has
decided to go another way, a way more suitable to NASes. You really don't
want one ifIndex per potential peer, do you (imagine a distributed, even
proxied RADIUS database of a large ISP) ? The Ascend way is logical for this
usage pattern: The box has one ifIndex per potential Interface and profiles,
when becoming active, are scheduled on these interfaces.
There is discussion about whether this still makes sense for _leased_
(aka nailed) lines and I understand that you may want to have a static
ifIndex for them. For dialed lines it's a no-no.
> "Oh, client ABC is currently using 45% of their capacity and since last monday
> they've used 4.4 gig."
I wouldn't do the latter with SNMP at all. SNMP is too unreliable for
that (if you want real numbers, no roundups). It is sufficient to just
get the picture. To get this information I would use either direct IP
accounting (net-acct, NeTraMet, inhouse solutions) or RADIUS accounting
or both. For illustrative purposes I use MRTG. Now MRTG suffers from the
same problem as your NMSs do: It is a fixed-interface-approach style hack.
I'm no perl hacker so I couldn't change MRTG directly. I have, however,
written a small Scotty script which extracts the correct ifIndex for a
certain profile from a Max and generates a mrtg.cfg style output with
Target-lines for MRTG. Works great for me. It has some shortcomings, mostly
it is a bit traffic intensive (to be compatible with 5.0A I must do a
full walk over the callStatusTable [some columns thereof, actually])
and does not work with Pipes (the session*Tables, while present, don't
list profile names correctly, still not fixed in 6.0.0). But it does
the job for me. My hopes are that MRTG 3.0 (said to be coming RSN) will
have direct support for Ascend in this way. If you're interested in the
script, see here:
http://www.ibh-dd.de/~beck/stuff/scotty/
A commercial NMS that should be able to do similar things ist NavisAccess
from Ascend.
--
Kanther-Line: PGP SSH IDEA MD5 GOST RIPE-MD160 3DES RSA FEAL32 RC4
+-o-+--------------------------------------------------------+-o-+
| o | \\\- Brain Inside -/// | o |
| o | ^^^^^^^^^^^^^^ | o |
| o | Andre' Beck (ABPSoft) beck@ibh-dd.de XLink PoP Dresden | o |
+-o-+--------------------------------------------------------+-o-+
++ Ascend Users Mailing List ++
To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd: <http://www.nealis.net/ascend/faq>
References: