Balancing the trade-offs of cloud management

How can cloud administrators manage the trade-off between overcommitting shared infrastructure and keeping users satisfied? This is a major issue for all cloud owners.

Overcommitting shared infrastructure is a basic part of the cloud model. It lets us do more work with fewer resources, thereby reducing the total cost of ownership for cloud providers. But that resource overcommitment can come with a cost, if not managed correctly. Overcommitting resources implies a risk of resource contention, a situation in which customers aren’t getting what they need, when they need it. To control customer dissatisfaction, cloud administrators need to balance overcommitment risks with overcommitment ratios.

In short, the overcommitment risk increases when the overcommitment ratio increases. Higher overcommitment ratios can reduce costs, but they increase the risk of congestion and customer dissatisfaction. On the other hand, overcommitment ratios that are smaller than necessary keep customers happy, but they also make cloud management inefficient and uneconomical.

The IBM Cloud Capacity Analyzer fills this gap. This first-of-a-kind tool from IBM Research provides recommendations for cost-efficient cloud capacity configurations that balance the trade-off between overcommitment risk, user experience and cost of configuration. A unique property of IBM Cloud Capacity Analyzer is that it reliably analyzes and helps plan capacity in the presence of the constantly changing population of virtual machines, coping with highly dynamic cloud environments. With Cloud Capacity Analyzer, infrastructure costs are reduced and user satisfaction is increased, bringing about improved productivity.

To learn more about IBM Cloud Capacity Analyzer, view the tool demo below:

What are your thoughts about balancing overcommitment risks and ratios? Leave a comment below.

————

About the Authors

Dr. David Breitgand is a research staff member at IBM Research – Haifa and has over fifteen years of experience in the areas of network, system and services management; fault-tolerant and distributed computing; and performance modeling and analysis. David is a technical leader of the Cloud Operating System Technologies group and an active contributor to the IBM Compute Cloud Reference Architecture.

Dr. Amir Epstein is a research staff member at IBM Research – Haifa. Amir is a member of the Cloud Operating System Technologies group. His research interests include cloud computing, approximation methods, online algorithms, algorithmic game theory, scheduling and load balancing.

Share
Comments: 3
David Breitgand

About David Breitgand

Dr. David Breitgand is a research staff member at IBM Research – Haifa and has over fifteen years of experience in the areas of network, system and services management; fault-tolerant and distributed computing; and performance modeling and analysis. David is a technical leader of the Cloud Operating System Technologies group and an active contributor to the IBM Compute Cloud Reference Architecture.
This entry was posted in Managing the Cloud and tagged , , , , , , , , . Bookmark the permalink.

3 Responses to Balancing the trade-offs of cloud management

  1. @CiprianoPF says:

    This understanding of Cloud Capacity Management is critical to truely gaining the full efficiencies and cost take-out of a Private Cloud deployment.

  2. I, too, see the issue that arises without fully managing a cloud. My only take on this is that this seems to be an age old issue just in different words; what's the difference between cloud management and more traditional hard drive management? We're still talking Gigs in the end.

Comments are closed.