I will reiterate my belief that I don't think its realistic for a simple
embedded device to maintain notification info for hundreds of clients. I
don't think its in the notification requirements document...maybe we
should talk about the number of notifications that an embedded device
should minimally support? And include the outcome of that discussion in
the requirements doc.
Also, I used my managed device examples (small hub to enterprise switch)
to illustrate low-cost to high-cost devices with very wide variances in
burden cost for manufacturing (memory, power, cpu, etc.).
Maybe what we are actually debating is semantics in how we define
"lightweight" and "heavyweight" Does lightweight mean simple to build
(overall)? Simple to code? Lightweight in terms of CPU cycles? Other? It
sometimes sounds like we are defining terms differently.
Randy
-----Original Message-----
From: imcdonal@eso.mc.xerox.com
[SMTP:imcdonal@eso.mc.xerox.com]
Sent: Sunday, March 22, 1998 4:37 PM
To: Joe_Filion@mc.xerox.com; ipp@pwg.org; jmp@pwg.org
Subject: IPP> Re: SNMPv3 unsuited for IPP/JMP
Notifications
Copies To: ipp@pwg.org, jmp@pwg.org, Joe_Filion@mc.xerox.com
Hi Randy, Sunday (22
March 1998)
My replies are imbedded in your note below, preceded by 'IEM>'.
I tried
to answer each of your questions.
Cheers,
- Ira McDonald (High North)
>----------------------------------------------------------------------<
>From: "Turner, Randy" <rturner@sharplabs.com>
>To: "'ipp@pwg.org'" <ipp@pwg.org>, "'jmp@pwg.org'"
<jmp@pwg.org>
>Subject: RE: IPP> Re: SNMPv3 unsuited for IPP/JMP
Notificiations
>Date: Sun, 22 Mar 1998 14:56:06 PST
>
>There two questions that have yet to be answered.
>
>You've repeatedly stated that SNMPv3 is not intended for "very
many
>notification clients". I would like to hear a technical
analysis of why
>you think this is the case.
IEM> Because the SNMPv3 security depends on the distribution of
shared
secrets to EVERY managed system, but does NOT specify a Key
Management
Protocol. Because the SNMPv3 Target and Notification MIBs have
highly
inefficient string indices, placing a significant burden on ALL
managed
systems. For ACTIVE management (not just passive monitoring) of
key
infrastructure (routers, switches, hubs, etc), SNMPv3 may turn
out to
be quite successful in the marketplace, but deployment will
undoubtedly
be sporadic and interworking problems will be common.
>Also, I think its VERY unrealistic that an embedded device will
have to
>manage "several hundred" client notification requests. If every
client
>can specify a totally different object set to be sent along
with each
>notification, then I don't think this is realistic for embedded
devices.
>If you have a defined trap/notification, with a specific list
of objects
>that are returned, then a notification originator only has to
keep track
>of transport domains and addresses, not every combination of
object sets
>that "possibly hundreds of notification clients" might want to
know
>about.
IEM> Xerox (like IBM, and various other printer vendors) builds
VERY big
production printing systems (which may cost over a million
dollars each).
The SNMPv1 framework defines 6 different important traps; each
interface
MIB (Ethernet, TokenRing, etc) defines several more traps; the
Printer MIB
defines one trap; the PWG Job Monitoring MIB will (hopefully)
define one
or several (job- vs document-level) traps; the XCMI (Xerox
Common Mgmt
Interface) suite of 14 MIBs defines 16 different traps. Xerox
clients
do NOT register promiscuously for ALL traps - they pick and
choose,
according to their end-user preferences or roles (operators and
system
administrators have different interests and needs, for example).
VERY
lightweight trap registration is a necessity for the printer
industry!
>I don't think SNMPv3 is heavyweight. It's designed be
implemented in
>everything from $300 managed 100Mbit hubs, all the way to
enterprise ATM
>switches. And I think whatever solution we come up with will
have to
>have the capability to support a distributed event notification
system,
>similar to the way SENSE and other distributed notification
mechanisms
>remove the burden from individual embedded printers from having
to keep
>up with the entire world of notifications.
IEM> But, your examples are all of infrastructure
(intermediate-systems),
not printers, scanners, and other end-systems. Managing a hub
or router
has essentially nothing in common with managing end-systems,
except that
they use a common protocol - SNMPv1, NOT SNMPv3, which is
unlikely to be
widely deployed for at least the next three years - look at the
deployment
track record of SNMPv2!
No vendor can tell their customers that they can't manage their
printers
until they deploy (largely non-existant) new enterprise open
management
systems (CA Unicenter, Tivoli TME, HP OpenView, Sun Solstice,
etc) that
have support for SNMPv3 (some of those that I just mentioned
don't have
support for SNMPv2, yet). Xerox research has found that many
customers
HATE to be told that "this product just requires a few little
server
applications." Elegant distributed event notification schemes
like
SENSE are very unpopular with Xerox product planners and
marketers
(and customers). Perhaps your customers haven't been bitten by
unstable
NetWare NLMs or NT server applications?
[Jay, I personally think SENSE is good stuff - but technical
excellence
isn't the only or even major factor in the deployment of
technology.]
>----------------------------------------------------------------------<