home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.protocols.tcp-ip.ibmpc:7354 comp.sys.novell:11346
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!cs.utexas.edu!hellgate.utah.edu!cc.usu.edu!jrd
- From: jrd@cc.usu.edu (Joe Doupnik)
- Newsgroups: comp.protocols.tcp-ip.ibmpc,comp.sys.novell
- Subject: Re: ODI information requested.
- Message-ID: <1993Jan11.155845.62631@cc.usu.edu>
- Date: 11 Jan 93 15:58:45 MDT
- References: <1993Jan9.085223.6519@osuunx.ucc.okstate.edu> <1993Jan11.200015.26324@novell.com> <1iskr9INNqfs@usenet.INS.CWRU.Edu>
- Organization: Utah State University
- Lines: 49
-
- In article <1iskr9INNqfs@usenet.INS.CWRU.Edu>, trier@slc6.ins.cwru.edu (Stephen C. Trier) writes:
- > In article <1993Jan11.200015.26324@novell.com> dmurphy@novell.com
- >(David Murphy) writes:
- >>This is the way it was designed and not a bug. The reasoning being
- >>that people would generally only being using a Stack for a certain form
- >>of connectivity, eg. TCP/IP from 1 company.
- >
- > I think it's more because it's hard to figure out just how to share one
- > Ethernet address between multiple protocol stacks. For example, just
- > how would you load more than one IPX for the same Ethernet address?
- >
- > The same applies to IP, though there has been a mechanism developed (in
- > NDIS) that permits multiple stacks of the same type, as long as they all
- > cooperate in a certain way.
- >
- > There's more than assumptions here. Most protocols are not designed for
- > more than one stack per Ethernet card, and that makes putting more than
- > one stack per card quite hard. Call it a protocol deficiency if you want;
- > that doesn't make it easier.
- >
- > --
- > Stephen Trier "We want to offer you a price that you
- > Network software type just can't afford to take advantage of."
- > Case Western Reserve University - Sales blurb from HSC Software
- > trier@ins.cwru.edu
- -------------
- Close, but not quite the story. Multiple protocols can share a board
- with little problem these days, and ODI, NDIS, and PD modules are designed
- for this. The difficulty is not sending packets of the same protocol type
- from multiple (same type) stacks. The difficulty is, however, receiving packets
- when there are contending receipients. A packet will be delivered just once,
- but to whom? Well, you can't tell from the outside so one stack gets the
- packet, whether it belongs there or not, and the others starve. Misdelivery
- has generally unpleasant conseqences. Delivering the same packet to all
- contenders is misdelivery to N-1 stacks, a worse problem than just one.
- A protocol stack is, for the most part, a decoder and encoder of the
- information in a packet, and that's a messy detailed business. ODI, NDIS, PD
- are not protocol stacks but are packet handlers of roughly similar nature;
- they are interested in physical addresses and the like rather than the deep
- interiors of what one protocol may say. They are good at separating one
- protocol flavor from another, because packets are built with this in mind,
- but can't distinguish IP for stack 1 from IP for stack 2 etc.
- Take this one step forward. UPD over IP. It's connectionless so no
- trail is left about who might want the incoming packet. The trail is an
- attempt to track a conversation, a la pktmux. Doesn't work with UPD, and
- in many cases it does not work with TCP.
- If you want multiple applications which use the same protocol then
- either run them all over the same protocol stack, or get another lan adapter.
- Joe D.
-