pesch@pa.reuter.COM (Roland Pesch) (07/12/90)
One of my colleagues, who currently has no access to the Net, is having
difficulty getting consistent answers from CISCO and IBM sales people to
the following question---perhaps the requisite knowledge is somewhere on
the Net?
(from Mike Symes, Reuters Toronto:)
We are proposing to connect an IBM MVS system to some LAN-based hosts
that run TCP/IP. The LAN's are accessible via an X.25 network, to which
they are connected by means of a CISCO router. Further, we are
considering connecting the MVS machine to the X.25 network by a 37xx
running NPSI.
BMC X.25 X.25 Ethernet
MVS: TCP/IP---37xx: NCP/NPSI---- net ----CISCO router-------other host: TCP/IP
We would like to know the following: will IBM's TCP/IP for MVS use the same
method as the CISCO router for transporting IP datagrams over X.25 virtual
calls? In other words, will the scheme work, or is it necessary to have a
mirror CISCO router at the mainframe end, connected to the mainframe in some
other way?
/Roland Pesch (pesch@pa.reuter.com)
Reuter:file Inc 425 Sherman Ave Ste 200 Palo Alto CA 94306satz@cisco.com (Greg Satz) (07/12/90)
>> We would like to know the following: will IBM's TCP/IP for MVS use the same >> method as the CISCO router for transporting IP datagrams over X.25 virtual >> calls? In other words, will the scheme work, or is it necessary to have a >> mirror CISCO router at the mainframe end, connected to the mainframe in some >> other way? I am not familiar with IBM's MVS TCP/IP implementation over X.25. If it implements RFC877 then it will interoperate with the cisco X.25 just fine. The only thing the IBM needs to do is place a call to the cisco's X.121 address with the first byte of Call User Data being 0xCC (hex CC). Then it needs to send the IP datagrams over the virtual circuit in individual X.25 packets. That is for each IP packet, there should be one X.25 DATA packet. For IP datagrams that are larger then the (possibly negotiated) X.25 packet size, the IP datagram will be sliced up into multiple X.25 packet size chunks and all but the last X.25 packet sent out with the M(ore)-bit set. Greg Satz cisco PS. Questions about cisco products can be sent to customer-service@cisco.com. We like mail.
pesch@pa.reuter.com (Roland Pesch) (07/13/90)
Thank you! That sounds much more specific than anything I believe
we've gotten so far, and in particular it should enable Mike Symes to
get matching specifics from the IBM end.
Unfortunately our mainframe guys don't connect to the Net much, so I'm
a human bridge in this conversation; BTW, I'm about to go on vacation
for a few weeks, so further communications may come from other human
bridges. Maybe we can even get Mike Symes interested in using one of
the systems near him with Net access.
/Roland