Comments on: Oracle clamps down on multicore licensing http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/ News and views from Pythian DBAs Fri, 10 Feb 2012 13:01:25 +0000 hourly 1 http://wordpress.org/?v=3.0.4 By: Marc Fielding http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-646255 Marc Fielding Tue, 22 Nov 2011 00:11:48 +0000 http://www.pythian.com/news/?p=1672#comment-646255 @Pieter, Yup, moving from 4-core to 64-core hardware would definitely increase your licensing costs. For this reason, licensing considerations often play a big role in hardware specification for Oracle servers. Could your client use a named user plus licensing metric rather than per-processor? @Pieter,

Yup, moving from 4-core to 64-core hardware would definitely increase your licensing costs. For this reason, licensing considerations often play a big role in hardware specification for Oracle servers.

Could your client use a named user plus licensing metric rather than per-processor?

]]> By: Pieter Viljoen http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-644447 Pieter Viljoen Sun, 20 Nov 2011 06:40:15 +0000 http://www.pythian.com/news/?p=1672#comment-644447 Just upgraded one site to HP DL980/G7 (64 Cores) from a Itanium (4 Cores). Not allowed to switch off the cores Not running extra jobs, just had to make the current ones run faster. Now interesting thing is that Oracle did not make the process go faster, Intel did. But Oracle wants gazillion more bucks for this achievement (which is not theirs to claim). Costs for Oracle is now totally unaffordable for the client. Don't like Enterprise DB (Postgres), but migrating client Just upgraded one site to HP DL980/G7 (64 Cores) from a Itanium (4 Cores).

Not allowed to switch off the cores
Not running extra jobs, just had to make the current ones run faster.

Now interesting thing is that Oracle did not make the process go faster, Intel did. But Oracle wants gazillion more bucks for this achievement (which is not theirs to claim).

Costs for Oracle is now totally unaffordable for the client.

Don’t like Enterprise DB (Postgres), but migrating client

]]> By: Ted Broadfield http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-360983 Ted Broadfield Thu, 28 May 2009 22:05:38 +0000 http://www.pythian.com/news/?p=1672#comment-360983 I came across a FAQ on Oracle's Partner Network portal under the pricing tab that deals with multi-core. One question addresses Standard Edition and Standard Edition One. It indicated that they will be counting cores with these lower database offerings. Amazing that they are going to increase pricing on their lowest end database offerings. I came across a FAQ on Oracle’s Partner Network portal under the pricing tab that deals with multi-core. One question addresses Standard Edition and Standard Edition One. It indicated that they will be counting cores with these lower database offerings. Amazing that they are going to increase pricing on their lowest end database offerings.

]]> By: Pythian Group - Blog http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-352210 Pythian Group - Blog Fri, 27 Mar 2009 17:17:59 +0000 http://www.pythian.com/news/?p=1672#comment-352210 [...] of like Oracle licensing. Pythian’s Marc Fielding detected Oracle clamping down on multicore licensing. In his analysis, “…future multicore server CPU generations will have a cost factor of [...] [...] of like Oracle licensing. Pythian’s Marc Fielding detected Oracle clamping down on multicore licensing. In his analysis, “…future multicore server CPU generations will have a cost factor of [...]

]]> By: Marc Fielding http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-351880 Marc Fielding Tue, 24 Mar 2009 19:03:18 +0000 http://www.pythian.com/news/?p=1672#comment-351880 Hi Tam, I was looking for a previous version of the table when writing the original post, and couldn't find one: archive.org doesn't archive pdf files, and google's cache has the new version. If anyone reading this has the old version though, please do send it on. Oracle likely changed the P6 factor for the same reason as the factor for Intel/AMD chips: to avoid giving specific advantage to one particular platform. As far as standard edition go, I can't see Oracle changing their per-socket licensing model because it would make them uncompetitive when compared to SQL Server. Hi Tam,

I was looking for a previous version of the table when writing the original post, and couldn’t find one: archive.org doesn’t archive pdf files, and google’s cache has the new version. If anyone reading this has the old version though, please do send it on.

Oracle likely changed the P6 factor for the same reason as the factor for Intel/AMD chips: to avoid giving specific advantage to one particular platform.

As far as standard edition go, I can’t see Oracle changing their per-socket licensing model because it would make them uncompetitive when compared to SQL Server.

]]> By: Tam Kyle http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-351870 Tam Kyle Tue, 24 Mar 2009 17:13:15 +0000 http://www.pythian.com/news/?p=1672#comment-351870 Hi Marc, the post prompted me to look further down the table - where it's now specified that AIX P6 chips have had similar limitations applied, ie these have a factor of 1.0 as opposed to the P5 factor of 0.75. I'm certain that this change in specification wasn't there in the previous version of the table. Does anyone have a hard copy of it for me to double check ? Hi Marc, the post prompted me to look further down the table – where it’s now specified that AIX P6 chips have had similar limitations applied, ie these have a factor of 1.0 as opposed to the P5 factor of 0.75.

I’m certain that this change in specification wasn’t there in the previous version of the table. Does anyone have a hard copy of it for me to double check ?

]]> By: Tam Kyle http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-351842 Tam Kyle Tue, 24 Mar 2009 11:15:40 +0000 http://www.pythian.com/news/?p=1672#comment-351842 Hi Marc, thanks for posting this - it's obviously further constraining the vertical scalability currently partially available with multi-core chips. It follows the IBM model somewhat of taxing the higher spec cpus with larger license requirements. I don't think it'll be long before Standard Edition gets a similar treatment ? Hi Marc, thanks for posting this – it’s obviously further constraining the vertical scalability currently partially available with multi-core chips. It follows the IBM model somewhat of taxing the higher spec cpus with larger license requirements.

I don’t think it’ll be long before Standard Edition gets a similar treatment ?

]]> By: Marc Fielding http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-351726 Marc Fielding Mon, 23 Mar 2009 17:46:45 +0000 http://www.pythian.com/news/?p=1672#comment-351726 Hi Joris, The big deal is that an existing Intel/AMD 4-core processor requires a 2-CPU Oracle license (4 cores * 0.5 factor). But all future processors (not just 8-core) will have a factor of 1, so that same 4-core processor in a future generation would require 4 CPU licenses instead of 2. An 8-core processor would require 8 full licenses ($332k with RAC). This will level the playing field with respect to SPARC, PowerPC and other factor-1 architectures, but for the growing Intel/AMD user population, it still means a 100% price increase. Marc Hi Joris,

The big deal is that an existing Intel/AMD 4-core processor requires a 2-CPU Oracle license (4 cores * 0.5 factor). But all future processors (not just 8-core) will have a factor of 1, so that same 4-core processor in a future generation would require 4 CPU licenses instead of 2. An 8-core processor would require 8 full licenses ($332k with RAC).

This will level the playing field with respect to SPARC, PowerPC and other factor-1 architectures, but for the growing Intel/AMD user population, it still means a 100% price increase.

Marc

]]> By: Joris http://www.pythian.com/news/1672/oracle-clamps-down-on-multicore-licensing/#comment-351685 Joris Mon, 23 Mar 2009 13:16:32 +0000 http://www.pythian.com/news/?p=1672#comment-351685 Hi, So far I understand the new multi-core table: http://www.oracle.com/corporate/contracts/library/processor-core-factor-table.pdf is that future generation CPU's will be counted as 1. So the next 8core will be calculated with multicore factor 1. I don't understand your post, because all current generation CPU's will be calculated with factor .5, as it is now. Greetings, Joris Hi,
So far I understand the new multi-core table: http://www.oracle.com/corporate/contracts/library/processor-core-factor-table.pdf

is that future generation CPU’s will be counted as 1. So the next 8core will be calculated with multicore factor 1.

I don’t understand your post, because all current generation CPU’s will be calculated with factor .5, as it is now.

Greetings,
Joris

]]>