Cloud Computing on Ulitzer

Last time we saw the difficulty in Predictive Rightsizing, a frustrating exercize based on "guesstimation" aimed at predicting future SLA in an everchanging business environment...

So what's the answer?

The solution is a truly dynamic, elastic, real-time on-demand SLA with a provisioning that is transparent to the users.

Provided that you stay within some reasonable boundaries, you can use as many resources as you need, or as few as required, ramping up and slowing down resource usage, without having to provision SLA in advance of usage.

This is the ultimate enabler for your business, because services are delivered at the exact level required, whenever needed, and without preoccupation of advanced provisioning.

This is called "pay per use", or "usage-based cost".

You pay for every network packet sent, generated or consumed, or for every Virtual CPU cycles used or for every record updated, and only for that. Not more, not less.

We are now beyond the nature of provisioning.

Figure 3 - Cloud Provider with pay-per-use business model.

The difficulty in this approach is to obtain crystal-clear visibility in the usage metrics; which means to obtain, store and analyze all the supporting data in order to get a precise count from which to generate the usage bills. Which means visibility in every network packet or CPU cycle of bytes used.  Using a single tool. In a non-intrusive way. Easy to deploy, easy to use.  Etc. etc...  Don't worry we'll see more about associated constraints.

So how would a Cloud Provider gain such hindsight on what is happening and how much to charge a customer?

Using logs of course.

Next, we'll talk about other challenges that are slowing down adoption of public Cloud Providers, Conflict of Interest and the notion of Trust, and Reversability.

And of course we'll deep dive on Log Management Best Practices for Better Clouds.

(I'd love to hear back from you, so fee free to drop me a line at   gorkasadowski (at) gorkasadowski (dot) com )