What really is pay-per-use in cloud?

One of the top benefits of cloud computing is pay-per-use. It’s included in many definitions as one of the key criteria for cloud. Cloud providers present it as their differentiator. But what are we really talking about? What does that mythical “pay-per-use” really mean? Before describing what is understood by the term, let me give a few examples.

  • When I use electricity I pay per-use. Actually to be completely correct, I pay in function of my consumption. I only pay for what I actually use.
  • If I place a call for example, I will pay for the duration of the actual conversation. I may have a different tariff rate depending on the type of call I place, but the rate is the same independently of the speed at which I speak or the amount of interactions between the two callers.
  • If I take a cab, things are different. I pay one tariff when the cab moves (and that is true pay-per-use) as I pay for each mile or kilometre driven. But when the cab stops at the red light, things become different. Now I pay another rate for the time I’m sitting there waiting.  
  • If I take a hotel room, I pay-per-reservation from the time I’m checking in till when I check out, regardless of how long I actually use the room.
  • If I’m a member of a gym, a golf or tennis club, I will most probably pay a monthly or yearly membership, regardless of how much I actually use the facilities.

Each of those examples is known to us. We find them normal and never discuss them. Things are actually similar in the cloud. I don’t why, but they are all collapsed in a single term, “pay-per-use,” making it often difficult to understand how things are really charged.

In my mind, there are four key payment mechanisms. I may be missing some, but let me explain the ones I think about:

  • Subscription based. A method found mainly in the SaaS world, where you have monthly or yearly fees to access the service. Your credentials are active for the subscription period and the system has an automatic or manual renewal. That depends on the service provider. Salesforce.com is a good example of such model.
  • Pay-per-reservation. A method where you pay from the moment you provision a service to the moment you de-provision the service. You remember my hotel example. It’s exactly the same and as it is the easiest way to implement billing in cloud systems, it’s not astonishing this method is often used. Amazon Web Services for example, implements such approach. How much you use the provisioned infrastructure is irrelevant. You will pay for the duration your virtual machine is available to you.
  • Pay-per-use. This is when I’m placing a call. The amount of time the service is in use is measured and a specific fee is applied. If I use the service at full capacity or only partially, it doesn’t make a difference. I pay for the time I use the service.

Consumption based. This is closest to the electricity example. My actual consumption is measured and charged to me depending on the amount of memory, CPU cycles, disk space, network traffic etc.,

I did not give any examples of the last two, because there aren’t that many of them yet. So, we call cloud billing as a pay-per-use model, but not many cases actually pay per use.

Now, we can be even more sophisticated. If we look at the true cost of a service, we will find that the cost drivers may differ. For example, the license of an operating system or a virtual machine is currently charged on a reservation base. The infrastructure on the other hand is pay-per-use or even better, consumption based. So, to have billing match the reality of the costs the closest, we could envisage a pricing schema that would combine consumption and reservation based pricing.

Who said “pay-per-use” is simple? It isn’t and covers many different approaches. Additionally the finer we measure, the more compute power we will require to measure. And this may affect the actual performance achieved. This is probably an area where “good enough” will do it.

Let me know what type of metering/charging you are looking for? Although this is not a scientific pole, it would be good to know what your take is on per-per-use in the cloud.  Which model do you prefer?

Labels: Cloud| cloud advisor
Leave a Comment

We encourage you to share your comments on this post. Comments are moderated and will be reviewed
and posted as promptly as possible during regular business hours

To ensure your comment is published, be sure to follow the Community Guidelines.

Be sure to enter a unique name. You can't reuse a name that's already in use.
Be sure to enter a unique email address. You can't reuse an email address that's already in use.
Type the characters you see in the picture above.Type the words you hear.
Search
Showing results for 
Search instead for 
Do you mean 
About the Author
Featured


Follow Us
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation.