Re: IBM Micro-partitioning and software licensing
Licensing question, not a technical bug: if a server supports micro-partitioning/LPARs, will vendors (including IBM for Informix) charge per fractional or partitioned CPU rather than for all physical CPUs in the box? Replies conflicted. One poster claimed an 8-way p570 running IDS in a 2-CPU LPAR would still need an 8-CPU licence; Art Kagel said his site is billed only for the physical CPUs its CPU VPs are bound to; Paul Watson noted it varies in practice by customer and relationship with IBM, some licensing all CPUs, some only the Informix ones. No definitive answer emerged, and the thread drifted into a joke about whether 'affine' is a verb.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
Neil Truby wrote: > <CKGeier@nwc.edu> wrote in message > news:1109621498.f3ba2a347ffe33e0d51595cbfa43822d@teranews... > >>This is a multipart message in MIME format. >>--=_alternative 0066729786256FB6_= >>Content-Type: text/plain; charset="US-ASCII" >> >>IBM has this really cool micro-partitioning setup in their newer mid-range >>servers where I can allocate 1/10 of a cpu's resources, for instance, to >>one program. Some companies products we purchase charge us maintenance and >>initial cost based on how many cpus we have in the server. Anyone >>purchased any products lately from companies who are willing to charge >>them for a fractional cpu, based on the micro-partitioning? (I would guess >>not). >>Thanks! >> >>Catherine Geier >>System Admin/IS support >>St. Paul, MN > > > Nor of course will IBM. If you have an 8-way p570 and you partition it up > and run IDS in one LPAR with, say, 2 cpus, they'll hit you for an 8-cpu > licence. > If I understand it properly! I dunno about that. We are charged by the number of physical CPUs we affine CPU VPs to. We run on 24-32 way Sun UltraSparc and normally use 2-8 processors for CPU VPs and that's the number of licenses we account for. Art S. Kagel
Art S. Kagel wrote: [cutting] >> >> Nor of course will IBM. If you have an 8-way p570 and you partition >> it up and run IDS in one LPAR with, say, 2 cpus, they'll hit you for >> an 8-cpu licence. >> If I understand it properly! > > > I dunno about that. We are charged by the number of physical CPUs we > affine > CPU VPs to. We run on 24-32 way Sun UltraSparc and normally use 2-8 > processors > for CPU VPs and that's the number of licenses we account for. > > Art S. Kagel > It seems to depend, I think the interpretion is more down to who you are and your relationship with IBM. We have clients on big CPU boxes some have full licenses even though they don't use them all the CPUs for informix , some only license the Informix CPUs -- Paul Watson # Oninit Ltd # Growing old is mandatory Tel: +44 1436 672201 # Growing up is optional Fax: +44 1436 678693 # Mob: +44 7818 003457 # www.oninit.com #
"Art S. Kagel" <kagel@bloomberg.net> wrote in message news:4225FCE7.2030309@bloomberg.net... > I dunno about that. We are charged by the number of physical CPUs we > affine > CPU VPs to. We run on 24-32 way Sun UltraSparc and normally use 2-8 > processors > for CPU VPs and that's the number of licenses we account for. First of all, only an American could mangle the noun "affinity" and come up with the verb "affine" :-) I don't have a definitive answer to my assertion about LPARS or VPARS yet, but you are definitely a special case here. No amount of pleading from normal, non-leading-news-agency customers about affinity processing or NUMCPUVPS settings will save ordinary mortals from being required to buy 32-cpu licences in the case you describe. Of course, this type of issue will generally only affect larger sites, of whom IBM is perhaps more likely to take more notice and maybe allow exceptions?
"Paul Watson" <paul@oninit.com> wrote in message news:d055hu$pe1$1@thorium.cix.co.uk... > Art S. Kagel wrote: > It seems to depend, I think the interpretion is more down to who you are > and your relationship with IBM. We have clients on big CPU boxes some > have full licenses even though they don't use them all the CPUs for > informix , some only license the Informix CPUs How much of this depends on whether the licences are supplied through PA?
Neil Truby wrote: > "Art S. Kagel" <kagel@bloomberg.net> wrote in message > news:4225FCE7.2030309@bloomberg.net... > > >>I dunno about that. We are charged by the number of physical CPUs we >>affine >>CPU VPs to. We run on 24-32 way Sun UltraSparc and normally use 2-8 >>processors >>for CPU VPs and that's the number of licenses we account for. > > > First of all, only an American could mangle the noun "affinity" and come up > with the verb "affine" :-) Guilty as charged. > I don't have a definitive answer to my assertion about LPARS or VPARS yet, > but you are definitely a special case here. No amount of pleading from > normal, non-leading-news-agency customers about affinity processing or > NUMCPUVPS settings will save ordinary mortals from being required to buy > 32-cpu licences in the case you describe. Mm. > Of course, this type of issue will generally only affect larger sites, of > whom IBM is perhaps more likely to take more notice and maybe allow > exceptions? It's not like we're cheating them out of anything. We're still paying WHOPPING numbers of $US for many hundreds of CPU licenses. Art S. Kagel
Neil Truby wrote: > "Art S. Kagel" <kagel@bloomberg.net> wrote in message > news:4225FCE7.2030309@bloomberg.net... > > First of all, only an American could mangle the noun "affinity" and come up > with the verb "affine" :-) There's a logic to it. If 'unity' is the noun, 'united' the adjective, and 'unite' the verb, then for the noun 'affinity' there SHOULD exist the verb 'affine' if there exists the adjective 'affined'! Voila: From The American Heritage Dictionary of the English Language: AFFINED adj. Linked by a close relationship. Beholden to another; bound. Which begs the equivalent definition: AFFINE v. To link by a close relationship. To create an affinity or objects which are affined. And as the final nail in the 'there ain't no such word' coffin: From Webster's Revised Unabridged Dictionary (1913) (web1913) Affine \\Af*fine"\\, v. t. [F. affiner to refine; ? (L. ad) + fin fine. See {Fine}.] To refine. [Obs.] --Holland. Art S. Kagel
Please get out more :-) Art S. Kagel wrote: > Neil Truby wrote: > >> "Art S. Kagel" <kagel@bloomberg.net> wrote in message >> news:4225FCE7.2030309@bloomberg.net... > > >> >> First of all, only an American could mangle the noun "affinity" and >> come up with the verb "affine" :-) > > > There's a logic to it. If 'unity' is the noun, 'united' the adjective, > and 'unite' the verb, then for the noun 'affinity' there SHOULD exist > the verb 'affine' if there exists the adjective 'affined'! Voila: > > From The American Heritage Dictionary of the English Language: > > AFFINED > adj. Linked by a close relationship. Beholden to another; bound. > > > Which begs the equivalent definition: > > AFFINE > v. To link by a close relationship. To create an affinity or objects > which are affined. > > And as the final nail in the 'there ain't no such word' coffin: > > From Webster's Revised Unabridged Dictionary (1913) (web1913) > Affine \\Af*fine"\\, v. t. [F. affiner to refine; ? (L. ad) + fin fine. > See {Fine}.] To refine. [Obs.] --Holland. > > Art S. Kagel -- Paul Watson # Oninit Ltd # Growing old is mandatory Tel: +44 1436 672201 # Growing up is optional Fax: +44 1436 678693 # Mob: +44 7818 003457 # www.oninit.com #