I could be wrong but I don't think many sites replace their mainframes
with "server farms". Normally they would replace them with big boxes,
rack servers or blades depending on the size of the system. HP
Superdomes used to be popular a decade ago but it seems Xeon class x86
servers can now do the job and you can easily build a large system with
sufficient failover for enterprise applications. Virtualization on x86
systems is so advanced these days that having to run huge amounts of
servers is a myth. They squeeze every last watt out of the boxes. I know
this because I've been working on projects moving data from z/OS to
Linux on x86 and have been doing sysadmin on the Linux boxes. Having
said that, migrating a complex application from a mainframe to any
alternate platform is a major risk and not one that most sites would
want to undertake.
On 4/05/2017 6:45 PM, Elardus Engelbrecht wrote:
James Wellingtin wrote:
I think that I have got a lot of good and fine answers from many people in this
discussion LIST You have really contributed in understanding of this topic. And
I also conclude from the amount of answers that it is something of great
interest ..
Just a few points from my side:
How do you plan to handle workload and response time on a server farm while
ensuring no deadlocks can take place?
DB2 and CICS have good measures to ensure no two records are updated at the
same time while keeping security and availability as top priority.
The customer run Z/OS 2.1 and CICS TS 5.x and DB2 version 11.
How many transactions are done per day and how many [active] users are working
there? Can the 'new' server farm' and all the network resources handle the
workload easily?
We have for CICS, for example, about 30 to 60+ million transactions per month, I don't
know how many are there for DB2 and other database systems. And we are "small".
I don't think it was mentioned in this thread, but a regular complaint I see is
- when the workload grows, you need to add more servers (with enough aircondit
and electrical feeds) and disks to handle the load while z/OS hardware has
already spare CPUs (and memory) available (if installed) waiting to be
activated as per Big Blue agreement.
Have anybody experience with the vendor MetaWare doing migration ,
Did they succeed in migrating. What did they migrate to How long time did it
take
Look at MetaWare long list of testimonials. Granted, you need to register
(which I did not, because I am not in the mood to receive spam e-mails), but
you probably will only see glowing testimonials and praises for MetaWare. I
wanted to see how big the clients own ex-z/OS systems were, but ...
Tools
What tools?
And of course , you may still come up with further experience , considerations
Ask them if they have something to replace IBM's Peer to Peer Remote Copy. I
believe there are similar things for that.
Ask them too whether they have products doing what HSM and SMS do?
Oh, ask them whether they can handle (on the server farm) Crypto and
transactions/sessions using SSL and friends. I know these servers can do that
in a way, but how do they that? And how fast (4K blocks per seconds for
example)?
Last question - ask them about backward compatibility.
Groete / Greetings
Elardus Engelbrecht
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
