Re: Processor Based License Model
From: John Vottero (John_at_mvpsi.com)
Date: 05/12/04
- Next message: david20_at_alpha2.mdx.ac.uk: "Re: You'll never guess what HP advertised"
- Previous message: Fabio Cardoso: "Re: TNSPING for OpenVMS Alpha"
- In reply to: Chuck Chopp: "Re: Processor Based License Model"
- Next in thread: Chuck Chopp: "Re: Processor Based License Model"
- Reply: Chuck Chopp: "Re: Processor Based License Model"
- Reply: John Smith: "Re: Processor Based License Model"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 12 May 2004 17:14:41 GMT
"Chuck Chopp" <ChuckChopp@rtfmcsi.com> wrote in message
news:Nxsoc.1523$U4.1279@bignews6.bellsouth.net...
> Barry Treahy, Jr. wrote:
>
> > I do know that for PC software, we're already had problems. We have
> > engineering design software that is ONLY available for Windows and they
> > crucify you with a major license increase if you run on multiple
> > processors. Well guess what the Intel P4 with HT looks like to Windows,
> > and to the software running on it? You guessed it!
> > I agree with you that the licensing models must be rethought, but I'm
> > sure that in the eyes of the CA's, Oracle's, HP's, etc., that would mean
> > they are being asked to 'leave money on the table' no matter how
> > short-sighted their perspective might actually be... I can't see them
> > making that choice quickly, or at all...
>
> And the really irritating aspect of a hyperthreading CPU appearing as a
dual
> CPU SMP system under Windows is that you don't necessarily get the same
> performance increase as compared to having 2 separate CPUs with one core
> each. If the on-board cache isn't increased and some of the other issues
> associated with memory access aren't resolved, you'll get a mult-core CPU
> that can't compete performance-wise with separate CPUs. Things like NUMA
> come to mind, along with instruction pre-fetching, pipelining, etc....
>
> Maybe for databases what we really need is a license that permits X number
> of transactions per seccond to be processed by the database. If you run
it
> on hardware that is insufficient to process the licensed TPS volume,
that's
> fine, you didn't under buy on your licensing and can perform a hardware
> upgrade w/o increasing your licensing costs for the database. But, if you
> do upgrade your hardware and the hardware could now have the database
> executing at a TPS rating higher than you are licensed for, the database
> would simply throttle itself back to your licensed TPS limit.
>
> Wouldn't this make more sense for database software licensing? It would
> certainly allow you to have a large SMP server and put a database
> application on it that doesn't monopolize the server's resources while not
> also paying through the nose to run that database application on that
large
> server.
It makes a lot of sense. Microsoft has already started doing something like
this. MSDE is actually SQL Server with a throttle and a db size limit of
2GB. The throttle kicks in at around 5 connections. You can redistribute
MSDE with your application (for free). Lots of applications will work just
fine with the throttle. If performance becomes a problem, you just buy a
SQL Server license.
Man would I like to have something like this for Rdb. Digital was close to
that when they bundled runtime Rdb with VMS. Selling Rdb was one of the
biggest mistakes Bob Palmer made.
>
> And, there's even other ways around the multiple CPU issue on Windows and
> Linux systems, too. Youd could implement server virtualization software
> like VMware [most likely GSX/ESX server] where each virtual machine only
has
> a single CPU regardless of how many CPUs the host system has. Then, you
> could distribute your database across those multiple virtual machines and
> run them in a "clustered" or "distributed" mode and thus bypass the
license
> fees associated with multiple CPUs per system.
>
- Next message: david20_at_alpha2.mdx.ac.uk: "Re: You'll never guess what HP advertised"
- Previous message: Fabio Cardoso: "Re: TNSPING for OpenVMS Alpha"
- In reply to: Chuck Chopp: "Re: Processor Based License Model"
- Next in thread: Chuck Chopp: "Re: Processor Based License Model"
- Reply: Chuck Chopp: "Re: Processor Based License Model"
- Reply: John Smith: "Re: Processor Based License Model"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|