This concept primarily applies to Public Cloud environments and should be considered when determining the ROI of any Cloud-based solution.
Morse Uncertainty Principle of Cloud Computing© (MUPCC)
The primary objective of business computing is to run applications that provide a benefit to the organization. MUPCC postulates that, in Cloud Computing environments, the amount of the environment being used specifically for the business application is adversely affected by variables that are dynamic and arguably uncontrollable. MUPCC has direct relevance to the ROI from Cloud Computing environments.
In Cloud Computing implementations there is a wide variety of software that is loaded and running on the application compute node. This software includes the Hypervisor, Operating System(s), General Management, Monitoring/Metering, Firewall, Antivirus, Backup, and others depending on the infrastructure design. MUPCC contends that the CPU, Disk, Network and Storage utilization of the combination of all the node software is sufficiently dynamic and unpredictable so as to adversely affect the primary application workload and reduce the business benefit expected from that workload relative to the investment in the environment.
When you add the effects of random workload migration between physical compute nodes in a virtualized environment, the adverse effects on the compute environment are increased because all workloads on both nodes – the sending node and the receiving node- are affected by the migration. CPU, Memory, Networking and Storage are consumed to perform the migration which affects all other workloads on the node. Further, depending on the overall computing environment, several, to many, nodes may be affected by inter-node workload migration because network bandwidth and storage are shared by some number of additional nodes.
The high level variables that affect each compute node include:
1. Primary business application demand
2. Number of installed Software Applications
3. Efficiency, quality and operational characteristics of all installed software
4. Virtual Machine density
5. Amount of node memory consumed at any given time
6. Amount of node CPU consumed at any given time
7. Amount of node-available network bandwidth consumed at any given time
8. Amount of data transferred in the storage environment at any given time
9. Number of VM node-to-node migrations at any given time
10. Policies and processes of the organization that affect the computing environment – backup schedule, VM snapshots, antivirus scans, patching schedule, etc.
11. Random Operating System behavior – indexing, disk maintenance, etc.
All of these variables directly affect the amount of compute resources that are available to the primary business application on a given cloud compute node.
Further, the infrastructure of the Public Cloud implementation could be sufficiently dynamic and unpredictable as to effect the creation of new computing instances simply because existing instances are resource constrained, precisely because of MUPCC. Organizations using Public Clouds need to investigate this concept to their satisfaction to ensure they are getting the resources they think they are and are getting billed according to the resources they use for their applications.
MUPCC should be considered when determining the ROI of a Cloud Computing infrastructure investment.
The term, “Morse Uncertainty Principle for Cloud Computing” is a copyright ©2011 of Paul Morse, Redmond Wa.