home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!gatech!paladin.american.edu!auvm!USCMVSA.BITNET!LDW
- Message-ID: <IBMTCP-L%92122222501581@PUCC.PRINCETON.EDU>
- Newsgroups: bit.listserv.ibmtcp-l
- Date: Tue, 22 Dec 1992 19:49:00 PST
- Sender: IBM TCP/IP List <IBMTCP-L@PUCC.BITNET>
- From: Leonard D Woren <LDW@USCMVSA.BITNET>
- Subject: MVS TCPIP problems (was Re: Usage of vector facility for TCP/IP?)
- Lines: 74
-
- > > "You're selling me the fastest general available computer on the
- > > planet and it connects to my network via this straw?"...
-
- And the straw is kinked, restricting how much you can force through it.
-
- > ...and that straw many times plain doesn't work. The current release
- > of TCPIP continues to offer unsurmountable impediments to the develop-
- > ment and deployment of client/server applications using concurrent
- > tcp/ip services...
-
- After a very discouraging phone call with TCPIP L2 and then with the
- TCPIP services manager, I will publicly say that I agree. For most of
- the problems that I've had open for months, I was told they're not
- going to be fixed, and that I can submit requirements. By the time
- MVS TCPIP is usable, MVS won't be around. Certainly not here. Lack
- of a decent MVS TCPIP will do more to kill off MVS than cheap unix
- workstations are doing. The future of computing (actually, it's here
- now) is networked computing. Any system that doesn't play in that
- networked world will die. Just because MultiTSO doesn't come from IBM
- is no justification for refusing to make IBM products work with it.
- If IBM had vision, TSO/E from IBM would look something like MultiTSO.
- MVS/DFP has just sent me a PTF to fix a very old VIO problem under
- MultiTSO. For years they refused to fix it. At least *they* have
- seen the light. But if TCP/IP isn't usable under MultiTSO, if you
- can't write multitasking servers, then that work will move elsewhere,
- to platforms where it does work.
-
-
- Here's a sampling of what I'm talking about:
-
- Problem: Simultaneous FTP/TELNET/any_VMCF_client in different Pie
- MultiTSO sessions results in address space termination with
- S40D.
- Response: "Submit a requirement."
-
- Problem: SMTP abends SB37 on incoming mail > 1.9MB due to hard-
- coded space value.
- Response: "Submit a requirement."
-
- Problem: IUCVMULT kludge doesn't allow multiple applications in
- different pie sessions.
- Response: "Submit a requirement."
-
- Problem: FTP TSO client doesn't honor dsprefix when looking for
- FTP.DATA.
- Response: "Submit a requirement."
-
-
- Note that these problems are *all* specific to the MVS version of
- TCP/IP. Most of them are difficult to fix due to the internal design
- of TCP/IP. The inability to reasonably do sockets applications under
- different TCBs in an address space is the show-stopper that Denis
- referred to above. The design of the whole IUCV interrupt mechanism
- is fatally flawed. It has taken me a while, but I now think that I
- know how to kludge it up some more to make it work. But IBM isn't
- interested, and since TCP/IP is a source-distributed OCO product, I
- don't think I can fix it myself.
-
- > it is an acknowledged problem and I am beginning to develop the
- > suspicion that IBM doesn't give a hoot about these defficiencies.
-
- Yep. They have "higher priority" problems that must be fixed. But
- the problem is that this multitasking issue never rises high enough on
- the list to be worked on seriously. A solution was proposed to me.
- When I pointed out why that solution wouldn't work, that seemed to be
- the end of it. I have been told in no uncertain terms that these
- problems are not going to be fixed in this release, if at all.
-
-
- I won't even comment on the performance problems which prompted this
- thread, except to say that the salespeople selling the big iron must
- be happy with the product.
-
- /Leonard
-