The History of Oracle Licensing Metrics

The History of Oracle Licence Metrics

Want to make more sense of Oracle Licensing Metrics? Learn how they evolved over time in this podcast and follow our latest in the series of Oracle licensing for beginners.

Metric mania

Oracle Licensing metrics. Photo courtesy of Josep Ma. Rosell(CC Attribution)

The Full transcript of the interview is below

Kay:                           Oracle licensing metrics are well-known for being complicated, and this is in part due to how they’ve evolved over time. Today I’m joined by Keith Dobbs, who’s going to talk through the history of Oracle licensing metrics. Hello, Keith.  

Keith:                       Hello there, Kay.  

Kay:                           Over to you, thank you.  

Keith:                       All right. A short story.  

                                    This is very much concentrating on the technology licencing, and if we think back to the old days when computers were servers with terminals– which is a pre-1999 world – Oracle introduced concurrent licensing for its servers. That’s where it was licencing the number of people accessing the server at any particular time. Not particularly worried about who connected to the server, providing it was in your own business, but just how many simultaneous connections there were. That worked well, very well indeed. 

                                    It worked well until the industry changed. In those days, there was one server, many terminals. But, bit by bit, the internet started to appear. At the end of say 1999, the internet was growing quite rapidly. It was getting to the stage where with concurrent device licensing you could have huge actual user numbers using very limited connections to a server. Which of course wasn’t much good for Oracle; it meant that more and more people were using the systems without paying for them, effectively. That was a point where Oracle decided it’s time to change licensing.  

                                    So it sat back and thought, rather than concurrent what can I have? There’s two kinds of licences required: one is for people who want to count users, and people who want to base their licencing on how powerful a server is. At that point, they introduced two new types of licence metric. The first was the Named User. But being Oracle it wasn’t going to be simple, so there were two flavors of Named User. There was a Named User single server licence, which allowed one user to use one server, or the more useful and higher cost Named User multi-server licence, which meant if you’re licensed by that one as a Named User, you can use multiple servers. You only have to be licensed once for your server estate. 

                                    For the server side of the licensing, they came up with a power unit. They thought well what’s the best way of measuring the power of a server? So for server-based licensing, they came up with a power unit which was based on the clock speed of the processor. Because the flavour of processors at the time were really going down two paths. There was the Intel route, I suppose in those days it was the Motorola and the Intel generating processors which were based on complex instruction sets. They came up with one power unit for those devices, which they called power unit CISC and they came up with another power unit called RISC which were the processors we know and love today, the ones which run UNIX –  the reduced instruction set computers. So they came up with two different pricing models for that. 

                                   It didn’t last for too long, because it started to get very complicated and they eventually changed it for what people probably know better, which is the universal power unit or the UPU.  

                                    As always with a Named User licence, Oracle wants a user minimum, the minimum named number of users you have to buy for a single server or whatever. The minimum was calculated as one user per thirty UPUs, where a UPU was the number of processors multiplied by the clock speed, measured in megahertz. So that was a nice simple way of working out user minimums as always.  

                                    However, that meant now that both the Named User and the server licencing was geared to the power of the machine. But of course, machine power was increasing terrifically, almost exponentially. So it didn’t take long before somebody who was licensed on a server, decided to refresh that server and realized that the next server they were going to get was probably ten times as powerful as the one they had before, which meant they’d need ten times as many UPUs or ten times as many Named User multi-server licences as they had before based on minimums. So the market required that something changed there. 

At that point, they introduced two new types of licence metric. The
first was the Named User. But being Oracle it wasn’t going to be simple,
so there were two flavors of Named User.

                                    This is really the stage where the NUMS, the NUSS and the UPU licences had to change and Oracle did two things. It brought in a Named User which had a minimum of ten users per processor, not dependent on the power of that processor, and they brought in the processor licence, which was simply licencing a processor irrespective of how powerful it was. I’m sure if you do the calculation you would find that meant prices went up as usual. 

                                    So now we have Named User and Processor, which is almost what we have today. By September 2002, the Named User evolved into what we know now which is a Named User Plus licence. The plus did a few things; one it increased the minimums from ten per processor to twenty-five per processor for Database Enterprise Edition. In a trade off against the increase in cost, Oracle improved and clarified licencing rules around data transfer backup and failover. So extra flexibility was built into that licence, or at least allowance for the way people tend to operate. 

                                    It’s this point where the much misunderstood “ten-day rule appears”, for fail-over. I won’t go into detail of that here,  we’ve other blogs which explain that better.  

                                    That continued through to July 2005, when processor core factors were introduced. A processor was no longer just a processor; processors now had cores and that meant more computer power in a processor and so Oracle edged a little bit back towards the power in a processor by saying that you had to take into account the number of cores multiplied by some core factor to work out the number of licensable processors in a server. 

                                    And that’s where it stands today, Kay.  

Kay:                           Okay, thank you very much, Keith for running through the progression of Oracle licencing metrics. Thank you.

So for some of you, I am sure you remember some of those older Oracle Licensing metrics and the pain in calculating Universal Power Units.

So why does it matter these days?  Well  believe it or not we do still come across client contracts that have concurrent licenses on them. Oracle is doing it’s best to transition everyone over to processors and Named User Plus metrics. If you are reviewed or audited and you present your contract paperwork with concurrent metrics be warned. They will challenge you on the ability to measure the usage. If in doubt you know who to call. Remember you can call the team at Madora in confidence to discuss anytime.


(2) comments

Add Your Reply