× The internal search function is temporarily non-functional. The current search engine is no longer viable and we are researching alternatives.
As a stop gap measure, we are using Google's custom search engine service.
If you know of an easy to use, open source, search engine ... please contact support@midrange.com.



VIOS slows down data transfer-- so true-- any interference in the transfer of data will slow things down.

However-- there are advantages on the VIOS side. We migrated to a V7000 several years ago. That first migration to external storage and VIOS 'interference' was no worse than other upgrades. However-- since then we've upgraded our CPU (I think twice) and migrated to a V9000.

Those last upgrades and migrations were marvelous! No more saving to tape (or other storage) and restoring, hoping and praying that everything went well. We upgraded the OS to match the new hardware, and *Magic* happened, migrating the OS and the data to the new hardware. If I remember correctly, the migration from internal disk to the V7000 happened without a save/restore as well-- and the V7000 to V9000-- the two were plugged together, and the data magically moved from old to new. At the magic moment, we shut down the old CPU, and booted the new-- and everything worked!

The extra overhead of the VIOS was compensated for by the higher power of the new hardware-- I know, it's a false improvement. But it's what has saved us from sloppy programming over the years-- users complain about response time, and we roll in a faster system.

There are other advantages-- our V9000 talks from our HQ to our 2nd V9000 at our DR site-- it's a link between the storage units; the CPU and OS/400 don't know anything about it.

We can copy our production LPAR to a 'flash copy' and save our entire system with a daily "Option/21" backup. Prior to the V7000/V9000, the only times we got an Option/21 backup was just before we migrated to new hardware, or upgraded the operating system.

So-- in spite of the extra overhead of using VIOS to run things, I think we've made a net gain.

--Paul E Musselman

.






At 11:28 PM +0000 7/26/20, Steinmetz, Paul via MIDRANGE-L wrote:
Jim,

Not sure where I read this, but IBM I internal disks (SSD) will out perform VIOS.
VIOS adds overhead and an extra level for the I/O.
Our Production LPAR runs on our host P9, 18 arms, 100% SSD
Our R&D LPAR runs as a client off our Production LPAR, 70 Virtual arms, 100% SSD.
R&D performance matches production, better at times.

Paul


-----Original Message-----
From: MIDRANGE-L <midrange-l-bounces@xxxxxxxxxxxxxxxxxx> On Behalf Of Jim Oberholtzer
Sent: Sunday, July 26, 2020 7:18 PM
To: Midrange Systems Technical Discussion <midrange-l@xxxxxxxxxxxxxxxxxx>
Subject: Re: IOPless machines and multiple CPUs

True, VIOS come with PowerVM, and is about the same cost per core as VMWare, or less.

That's the software. Now get the fiber And Ethernet switches to support the virtual environment and the cost goes up dramatically, so does the complexity.

For true production workloads that have any sensitivity to I/O, or the performance is critical, then VIOS is the way to go. Many IBM i workloads are just fine with IBM i hosting, and that's far less complex, simple to set up, and easy to maintain.



Jim Oberholtzer
Agile Technology Architects



On Jul 26, 2020, at 12:01 PM, Roberto José Etcheverry Romero <yggdrasil.raiker@xxxxxxxxx> wrote:

¤Configuring an LPAR with more cores than licensed ends up with a
message saying that you are over entitlement but it still uses those cores.
One of the reasons I don't understand i hosting i is the economics of
the exercise. It is a lot more expensive to use IBM i to virtualize
storage/network than the almost free VIOS.
On a big machine you might need more than 2 entire cores for IO and
that gets expensive quickly.

Roberto

On Sun, 26 Jul 2020, 12:39 Patrik Schindler, <poc@xxxxxxxxxx> wrote:

Hello,

imagine a newer POWER machine with, say, four CPUs. IBM i is licensed
to one CPU, so I guess when you give the LPAR more than one, it won't
be used for running "normal" code. Right?

Does anybody know (and provide evidence) if a second assigned CPU
>> will be used as "helper CPU" for I/O? Not exactly an IOP, since this
involves drivers and other stuff. My thinking stems from the z-world
where processors can be freely configured to specialized tasks (I/O,
running Linux, running Java, ?). The number of CPUs often influences
licensing cost, which probably was one of the reasons to invent this
feature in the first place.

Thanks!

:wq! PoC

PGP-Key: DDD3 4ABF 6413 38DE - https://www.pocnet.net/poc-key.asc
>>

As an Amazon Associate we earn from qualifying purchases.

This thread ...

Follow-Ups:
Replies:

Follow On AppleNews
Return to Archive home page | Return to MIDRANGE.COM home page

This mailing list archive is Copyright 1997-2024 by midrange.com and David Gibbs as a compilation work. Use of the archive is restricted to research of a business or technical nature. Any other uses are prohibited. Full details are available on our policy page. If you have questions about this, please contact [javascript protected email address].

Operating expenses for this site are earned using the Amazon Associate program and Google Adsense.