It’s been three years since the National Institute of Standards and Technology (NIST) released its final definition of cloud computing ─ after developing 15 drafts over a couple of years. The definition didn’t change noticeably from its initial iterations and today still includes the following characteristics: on-demand self-service, broad network access, resource pooling, rapid elasticity and to be a measured service.
The Benefits of On-demand Resources
That first characteristic ─ on-demand self-service ─ was a new concept for many IT organizations as it removed the traditional barriers between users’ needs for IT resources and their delivery. A cloud consumer didn’t need to interact with internal IT staff or a cloud services provider (CSP). Instead, a catalog of available IT resources could be directly accessed via a self-service portal, allowing the user to provision what was needed when it was needed.
It was a great model. Users could access the IT resources they needed, such as a new virtual machine (VM), without the delays inherent in having to make the provisioning requests, wait for them to be approved and then wait even longer for the cloud administrators to manually provision the requested resources. Self-provisioning also took pressure off IT, sparing them some of the labor involved in fulfilling users’ service requests.
The Problem with Creep
A lot of CSPs jumped on this, incorporating self-service tools into their cloud offerings. In their rush to make self-provisioning available to users, however, many overlooked some of the downsides ─ most notably the issue of “creep”.
On-demand self-provisioning is supposed to help users benefit from one of the other NIST-defined characteristics of the cloud: rapid elasticity. The problem is that too often the elasticity only stretches one way ─ out. Users keep helping themselves to more resources, often whether they need them or not. This wouldn’t be a problem if the resources were free or were being well utilized. But, they aren’t.
This hoarding causes escalating costs and resource underutilization. At some point capacity limits can also be reached, resulting in a scarcity of resources, lost agility and wasted capital, power and cooling.
The problem is that self-service provisioning, at least as offered by many CSPS, tends to work like an all-you-can-eat buffet. Users keep coming back for more – even if they haven’t eaten (or in the case of IT resources, used) what they already took. There are no accountability requirements and no automated process to identify or reclaim unused or excess resources. And unlike the all-you-can-eat buffets, there are charges for all those excess resources taken.
Rightsizing for Optimized Resource Usage
That’s why the self-provisioning model offered by Peak 10 is so appealing. It includes built-in rightsizing recommendations to help users optimize resource consumption. Cloud users can define limits, and the system will let them know when they are running out of resources and even justify additional hardware purchases.
Specifically, Peak 10’s self-provisioning capability entails resource consumption trending to provide high-level management reports to show how resources are being consumed and who is consuming them. Resource monitoring, rightsizing and reclamation features help identify and reclaim unused resources and eliminate VM sprawl.
This right-sizing capability makes the Peak 10 cloud a great option for businesses want cloud services delivered their way ─ and with the information to make sure that their way is the most efficient, economical way.