RE: Non-CDB officially deprecated ...

From: Mark W. Farnham <mwf_at_rsiz.com>
Date: Fri, 23 Jan 2015 07:05:49 -0500
Message-ID: <2f9301d03704$ee26c720$ca745560$_at_rsiz.com>



“What’s the point?”  

My presumption is the point is not having to maintain two code paths and, probably more importantly, two support knowledge variables in the long haul.  

I would not look for direct customer benefits in removing this choice, only the indirect benefits of it allowing Oracle to be more efficient in development and support.  

So far this is only deprecation and there is no upcharge for a single-tenant CDB/PDB configuration. Only IF some per host or RAC set of hosts requirement creeps in such that you cannot have multiple single tenant CDB database per host or RAC set of hosts would it really be a license issue, and only then if it were actually de-supported. De-support cannot occur without de-support of any items that remain at that time on the list of features (and any applications owned by Oracle that require any of those features) that the CDB/PDB architecture does not (yet) support.  

If you have very many databases that are difficult to consolidate into a single database for one reason or another there is theoretically a solution space where the aggregate of the overhead running them on a single large machine as independent databases or some variable grouping of the independent databases on multiple machines is more expensive than some set of machines some of which are configured as multi-tenant databases. Identifying an actual solution space would involve considering the full range of options as you discuss below for a specific user of Oracle technology as long as the license upcharge for the multi-tenant part of the configuration remains. Since it is an overall cost calculation, whether or not there is a solution space in favor of multi-tenant would be dependent on license fee and support fee negotiations.  

Oracle’s push to be competitive in the two socket market space will likely play a role in future license cost policies.  

mwf  

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Mladen Gogala (Redacted sender "mgogala_at_yahoo.com" for DMARC) Sent: Friday, January 23, 2015 1:56 AM
To: oracle-l_at_freelists.org
Subject: Re: Non-CDB officially deprecated ...  

On 01/23/2015 01:16 AM, Chitale, Hemant K wrote:

A CDB with a *single* PDB (aka “Single Tenant”) does not require the MultiTenant Option. It is implicitly in the Enterprise Edition.  

Hemant K Chitale    

What's the point? Why would I want to complicate my life with CDB's and PDB's, without getting any benefit? Jared's poll doesn't show overwhelming enthusiasm for upgrading to 12c in the first place. I believe that people are still waiting for 12.2. Making upgrades more complex and requiring them to learn new paradigm, without having any benefits from that would only postpone upgrading en mass. I do believe that multi-tenant option will be integrated for free in the future releases because from financial point of view, it simply doesn't make sense. How would you justify the purchase of PDB's to your boss? Say you have a big Linux box with 256GB RAM and 32 CPU cores. You are running 3 instances on that machine. What would you say? I doubt that something like: "for only quarter of a million, I can save 15% or more of CPU power and 32 GB RAM" would fly. There are many other options: use NUMA server and tie each instance to the particular set of CPU resources. You can also virtualize DB servers and have the servers running on the separate virtual machines. With some investment in the hardware, the same effect will be achieved at a fraction of the price. This announcement doesn't make any sense whatsoever, unless Oracle Corp. is planning to integrate PDB's in the future releases for free. I cannot envision them selling too many multi-tenant licenses right now. It just doesn't make sense from the financial point of view. Of course, making that public would also prevent them from selling any multi-tenant licenses right now because the clients would simply wait for 12.2. Which they're doing right now, according to Jared's poll.

--

Mladen Gogala
Oracle DBA
http://mgogala.freehostia.com

--

http://www.freelists.org/webpage/oracle-l Received on Fri Jan 23 2015 - 13:05:49 CET

Original text of this message