At 06:33 PM 4/15/01, Chuck Larrieu wrote:
>I've heard the same said of IPX.
Yup, people who say IPX is "chatty," are usually blowing smoke too. ;-)
From a very quick assessment of IPX, I would say that it's pretty well
behaved:
SAP broadcasts every 60 seconds can get out of hand, though.
Use SAP filters to avoid too much traffic.
IPX RIP sends every 60 seconds, much better than RTMP's 10-second timer.
NLSP or EIGRP are alternatives to IPX RIP and SAP.
NCP keepalives are even worse than ATP keepalives. They are every 5
seconds. (Use Cisco watchdog feature to avoid this over DDR lines.)
IPX sends broadcasts, rather than multicasts, which is kind of rude.
There is no ARP as you mentioned, which is a good thing from a traffic
point of view.
>Which leads to the obvious question - is IP
>any less chatty than any other protocol?
IP was designed with different goals in mind. AppleTalk and IPX were
designed to be easy on the user. A tradeoff is that things like service
location are easy, but they cause extra traffic. IP has more options for
routing and the newer options such as OSPF are not chatty. For file
sharing, you would have to use NFS or SMB over NetBIOS over UDP/IP, which
can be chatty, depending on the configuration.
Microsoft Networking can be very broadcast-intensive if NetBEUI is used.
But most people don't use NetBEUI anymore.
Well, I wish I had more time to discuss this interesting topic, but I
better get back to work.
Priscilla
> At least with IPX there is no need
>for ARP because the station address and the MAC address are one in the same.
>
>I believe that the designers of Microsoft networking were cognizant of the
>broadcast issue, which is why there are such long periods in that protocol's
>update cycle.
>Chuck
>
>-----Original Message-----
>From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of
>Priscilla Oppenheimer
>Sent: Sunday, April 15, 2001 6:47 PM
>To: [EMAIL PROTECTED]
>Subject: Re: Packet retransmission [7:662]
>
>At 04:04 PM 4/15/01, you wrote:
> >Priscilla,
> >
> >With reference to the comment below:
> >
> >The cram
> > > session has the usual misconceptions, such as claiming that SRB and SRT
>are
> > > in 802.5, which they aren't, and that HSRP is a routing protocol that
is
> > > standing by, (it's a router standing by), and AppleTalk is "chatty."
> > >
> >I have two questions:
> >
> >1) Has the SRB definition only recently moved to IEEE 802.1d?
>
>SRB is documented in IBM's Token Ring Architecture Reference Manual. It's
>not in an IEEE document. When IBM brought the SRB specs to the IEEE, the
>IEEE said that SRB must fit with existing bridging standards, and IBM and
>IEEE jointly developed SRT and added it to 802.1D in the early 1990s.
>
>One area of confusion is that SRT was designed mostly by people on the
>802.5 committee, including IBM engineers. Some preliminary documents said
>802.5 on them. But the intent was that it be part of 802.1D right from the
>beginning. (I was on the 802.5 committee for a short time in the early
>1990s and that's the impression I got anyway.) It's a picky thing, but I
>like to point it out in case someone actually wants to ready the
>specifications.
>
> > I do not
> >have the very latest versions of both IEEE 802.5 and 802.1d but keep
> >thinking (probably wrongly) that SRB is included in the former one. At
> >least in previous published version I have got. I know the SRT is 802.1d
> >but cannot find any indication in the standards and status reports on
> >IEEE concerning SRB.
> >
> >2) I also though the AppleTalk is quite chatty. Chooser is given as a
> >usual example for that. Or is it only when the Chooser window is left
> >open unnecessarily?
>
>Apple fixed the excessive traffic caused by leaving the Chooser open in
>1989. (System 7). It was only a lot of traffic if the user had also
>highlighted an object type (printer, server) and zone name. It was never
>really a serious problem. Usually the user didn't have those highlighted.
>
>There are a couple legitimate reasons to call AppleTalk "chatty." The
>10-second timer for RTMP is awfully small. But the advantage to a small
>timer is quick convergence. Also, end stations learn very quickly who their
>new router is. There's no need for HSRP. The other case where AppleTalk is
>chatty is the AppleTalk Transaction Protocol (ATP) sends keepalives to the
>other side every 10 seconds.
>
>But, AppleTalk does not advertise zones. (If you see that common
>misconception in a book, throw it out. ;-) The request to learn the zones
>associated with a network is a unicast frame. AppleTalk does not broadcast.
>It multicasts. A well-behaved NIC in a PC should not bother the PC CPU with
>AppleTalk multicasts.
>
>I bristle when I see documents that are clearly oversimplified making the
>statement that AppleTalk is chatty. If you're just going to say a few
>things about AppleTalk, you could mention the good things: easy
>configuration, dynamic addressing, easy resource location, etc.
>
>Priscilla
>
>
> >Rita
>
>
>________________________
>
>Priscilla Oppenheimer
>http://www.priscilla.com
>FAQ, list archives, and subscription info:
>http://www.groupstudy.com/list/cisco.html
>Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
________________________
Priscilla Oppenheimer
http://www.priscilla.com
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=833&t=662
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
