I've had the same discussions with networking groups at different
clients. One thing for sure: networking procedures and plans remain
opaque to me. Anway after months of discussion one client's network
group has agreed to go with a vswitch solution. They are massively happy
with it. Redundancy is provided through trunked real switches attaching
to two different osa features. The tcpip vswitch machines have multiple
sets of devices. Works brilliantly!
Next big play I have in store is to go vlan big time. The current
network architecture is just stepping up to vlan. Vlan would save the
client a whole bunch of OSA devices. Probably take 'bout a year.
Keep at it as it all we can do!
David
PS. also like to do some in board firewalling - may have an IFL ready
just for fw'ing! But that's another story
Michael Coffin wrote:
Hi David,
Yes, that's correct (i.e. no VIPA). All of the TCPIP subsystems are on the same
network, no VSWITCH, VIPA or automatic-failover between them. I wanted to use
VIPA, but let's just say the people that own the network at this site "weren't
getting it". They had a LOT of trouble understanding how our Linux/390 guests
could have no "real NICs" with MACs, cables, proxy-arping, and the like - or the
fact that we had multiple TCPIP "servers" all running on the same physical piece
of hardware. I was lucky to get reserved IP addresses for my TCPIP's and
Linux/390's - had to "create" some MAC addresses for the Linux/390 guests just
so they'd have something to put on their "request form". :)
I spent months trying to get the network resources I needed to deploy VIPA - but
finally gave up. Maybe one of these days I'll revisit it - but it was mighty
painful the first time around. :)
PS: We use a BusTech NetShuttle with 3 GigE paths, 2 100mbps paths (primarily
for backup and development only) all over one ESCON connection to our MP3000.
One GigE is dedicated to a TCPIP server used strictly for Linux/390 traffic, the
other 2 GigE's are dedicated to the primary TCPIP server for interactive CMS use
(along with supported services like printing, spooling, file transfer/FTP, etc.,
etc.).
Michael Coffin, President
MC Consulting Company, Inc.
PMB 123
289 Park Street
Stoughton, Massachusetts 02072
Voice: (781) 344-9837 FAX: (781) 344-7683
[EMAIL PROTECTED]
www.mccci.com
We employ aggressive SPAM filters. If you cannot reply or send email to
mccci.com go to www.mccci.com/spamblockremove.php
-----Original Message-----
From: VM/ESA and z/VM Discussions [mailto:[EMAIL PROTECTED] On Behalf
Of David Kreuter
Sent: Wednesday, January 04, 2006 10:43 PM
To: [email protected]
Subject: Re: TCPIP stacks
Hi Mike: By backup, you mean the stacks all connect on the same networks but are
not providing hot backup through VIPA or VSWITCH solutions, right? So that
TCPIPA et al doesn't route to TCPIPB et al? David
-----Original Message-----
From: VM/ESA and z/VM Discussions on behalf of Michael Coffin
Sent: Wed 1/4/2006 1:26 PM
To: [email protected]
Subject: Re: TCPIP stacks
FWIW - I use an entirely separate second "TCPIP Subsystem", complete with it's
own TCPIP server. So I've got TCPIP (my primary TCPIP subsystem with it's
stack), TCIPIPA and TCPIPB - all running completely independent of one another,
all with their own hardware, IP addressing and service virtual machines (i.e.
TCPIPA has ROUTEDA, NAMEDA; TCPIPB has ROUTEDB, NAMEDB, etc. etc.).
Not only can I bounce the target TCPIP virtual machine without it impacting
other TCPIP subsystems, but if one fails entirely - I have the others as backup.
This is particularly helpful when, for example, you need to bounce your primary
TCPIP virtual machine and are at a remote location - just connect using TCPIPA
or TCPIPB so that your TN3270 session won't disappear when the primary TCPIP
virtual machine is taken down.
Not a lot of work IMHO. :)
Michael Coffin, President
MC Consulting Company, Inc.
PMB 123
289 Park Street
Stoughton, Massachusetts 02072
Voice: (781) 344-9837 FAX: (781) 344-7683
[EMAIL PROTECTED]
www.mccci.com
We employ aggressive SPAM filters. If you cannot reply or send email to
mccci.com go to www.mccci.com/spamblockremove.php
-----Original Message-----
From: VM/ESA and z/VM Discussions [mailto:[EMAIL PROTECTED] On Behalf
Of Thomas Kern
Sent: Wednesday, January 04, 2006 1:08 PM
To: [email protected]
Subject: Re: TCPIP stacks
I use a second stack on both sides of a z890 for TELNET and some specialized
services across a HiperSocket. It took about 30 minutes to make the necessary
updates to the appropriate DTCPARMS and to create the directory entries.
Creating the full suite of application servers to go with a second stack would
be much more work and I would not want to do it without a great deal of
justification.
/Tom Kern
/301-903-2211
