cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

RHEL for Virtual Datacenters Points rule set

Greetings!

The attached RHEL custom points table is a puzzle to me, anyone understand how the number works?

It seems that when you set a RHEL license to license type Processor Points and set Points rule to RHEL for Virtual Datacenters, this is the rule to consume your license.  

I can understand that a On-Prem 2-socket server consumes 1 Point, but the rest of numbers do not make sense to me at all. Like why a 4-socket and 8-socket server each consumes 1 Point as well.

I would appreciate if anybody can shed some light on this. Thank you so much!

 

(1) Solution
nrousseau1
By Level 10 Champion
Level 10 Champion

Hello @LChen2216b5 ,

The point rule for RHEL for DataCenter looks complex indeed, but taking example, to will hopefully mahke sense. First, important thing, when you import you RHEL subscriptions, you should use a quantity per unit of 2. Indeed, each license you purchase allows to cover 2 processors. With a Quantity per unit of to, you have a count of the "number of licensed processors". Then, the License will count a "Number of Processors to license".

The second challenge is that you need to "round up to the next even number" the number of processors to license.  Indeed, a 3 processor server needs 4 licenses, a 5 need 6 licenses and so on.

The weird factor just intend to round up the number of processors for the servers. For instance, for a 7 processor server, you need 7 * 1.1428 =  8 licenses.

I hope it makes sense,

Best regards,

Nicolas

Product Manager - ITAM

View solution in original post

(6) Replies
nrousseau1
By Level 10 Champion
Level 10 Champion

Hello @LChen2216b5 ,

The point rule for RHEL for DataCenter looks complex indeed, but taking example, to will hopefully mahke sense. First, important thing, when you import you RHEL subscriptions, you should use a quantity per unit of 2. Indeed, each license you purchase allows to cover 2 processors. With a Quantity per unit of to, you have a count of the "number of licensed processors". Then, the License will count a "Number of Processors to license".

The second challenge is that you need to "round up to the next even number" the number of processors to license.  Indeed, a 3 processor server needs 4 licenses, a 5 need 6 licenses and so on.

The weird factor just intend to round up the number of processors for the servers. For instance, for a 7 processor server, you need 7 * 1.1428 =  8 licenses.

I hope it makes sense,

Best regards,

Nicolas

Product Manager - ITAM

Hi Nicolas,

Thank you so much for your help!

If I understand your words correctly, if I set the Qty per Unit to 2 for a RHEL license, then the license has 2 points. For a server with an even-number, say 2n processors, the server should be covered by n licenses, each processor consumes 1 point; for a server with an odd-number, say 2n+1 processors, the server should also be covered by n+1 licenses, each processor consumes (2n+2)/(2n+1) points. 

If I wish to produce a compliance report for RHEL for Virtual Datacenter licenses, I should compare the number of points consumed against the number of points purchased.

Thanks again and you have a wonderful day!

Lily

Exactly, we compare processors (purchased so, The RHEL Purchases should have a quantity per unit of two.) to the consumed processors (that must be even because you can't consume half of a license). I would advise to use the Red Hat Linux Server Optimization Report  (released in 2021R1.1) to understand the "low density" and "high density" clusters (that are worth convering with DataCenter node edition). The following SAM Best Practice Webinar provides more information on the Red Hat Optimization report. This one shows how to us the Intelligent license restriction to "focus" licenses on optimal clusters.

All SAM Best Practices webinars can be found in the SAM Best Practice Hub

Best regards

Thanks Nicolas. The extra information you provided is really helpful.       

While we're at it, I wonder if I could bother you with a quick follow-up question: Is it possible that Flexera has inconsistent implication regarding "Point"?                                                                                                                                                   The reason why I ask is that if I look at the Points Rule Set for RHEL for Virtual Datacenters, I think “Point” here acts something like a multiplier to the number of sockets/processors of the server.  If I take a 5-socket sever as an example, it will consume 5*1.2 = 6 points (3 licenses); While "Point" in the Optimization report is Synonym for license —I looked at a random cluster with 2 processors (System suggests the Optimal license is Red Hat Enterprise Linux for Virtual Datacenters because we have 14 VMs on host), the consumed point I get from the report is 1. It should be 2 points (1 license) if we adhere to the previous "Point" definition, right?

OR it could be that there is something I got it wrong that I'm not aware of.

Best Regards,

Lily

 

 

 

That's right, the license approach is a "Processor Point (Each license has two process points)" and the Report shows Licenses (with eventually some fractional consumptions for Server node when hosts have uneven number of VMs). This is a little bit like Windows Server where the "processor" rule was computing licenses consumption (qty per unit was one) and the new core rule (introduced in 2016) is count cores (so, the Qty per unit must be one). I would say in the end, both report and licenses are right (and the report computes the optimal license per cluster) and the "inconsistency" is not impacting the license consumptions (once you have in mind what you consider a "License Right".

The report can show fractional consumptions, not the license, that's why we could do this "license consumed" approach in the report. Using the reports recommendations with make the license count the right consumptions. Hope it helps.

Nicolas

Thanks Nicolas. I appreciate your time and guidance.

Best Regards,

Lily