Browse Prior Art Database

Real-Time Predictive Cloud Optimization Disclosure Number: IPCOM000244735D
Publication Date: 2016-Jan-06
Document File: 5 page(s) / 563K

Publishing Venue

The Prior Art Database


Disclosed is a of method and system for making the cloud smarter by using predictive analytics and real-time optimization, while ensuring that customer needs for infrastructure, platforms, and software can be met using an available set of platforms using an approach that optimizes the overall customer experience and satisfaction level with the cloud provider.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 43% of the total text.

Page 01 of 5


A method or system is needed that enables system administrators to anticipate problems in the cloud prior to occurrence, and take pro-active actions in real-time to prevent abnormal conditions in the cloud (e.g., exceeding capacity, exceeding hardware and energy budgets, failures due to cyber-attacks, etc.). To address business and technical optimization problems a smarter cloud enabled with real-time Big Data

Analytics is needed.

The novel contribution is a system for real-time predictive cloud optimization and capacity planning. The method applies big data analytics to provide smarter cloud management and optimization in real-time.

Figure 1: Cloud Control Center

Use Case 1: Real-Time Predictive Cloud Optimization and Capacity Planning

In this use case, the objective is to use predictive analytics and real-time actions/decisions to optimize the physical layer of the cloud and the associated energy costs by minimizing the number of physical hosts and maintaining a balanced workload across the active hosts through real-time VM migration. The approach uses historical data to train the predictive models and processes monitoring data to anticipate potential problems and run optimization analytics to take actions in real-time before a potential problem happens. For example, the analytics attempt to anticipate situations in which the cloud system reaches the capacity limit of a physical host (in terms of central processing unit (CPU), memory, disk, network, etc.), and situations in which resources are not used optimally because of poor mapping between hosts and VMs. The objectives of the optimization analytics include the following:

• Minimizing the number of physical hosts (for cost and energy saving)

• Minimizing the standard deviation of the VMs activities across active hosts (Spreading/balancing the VMs loads across active hosts)


-Time Predictive Cloud Optimization

Time Predictive Cloud Optimization

Page 02 of 5

• Minimizing the number of VM moves/migrations

Use Case 2: Cloud Resiliency: Real-Time Preventive Measures against Cyber Attacks

In this use case, the objective is to use real-time analytics to detect or predict patterns that can characterize cyber-attacks (e.g., simultaneous/synchronized and highly correlated activities in a number of VMs that are supposed to be unrelated to each other). Then, administrators and systems can take preventive measures in real-time before the attacks grow or spread to other VMs/hosts (e.g., taking affected VMs off the network, changing Internet protocol (IP) addresses, load migration, etc.).

For both use cases, the broad steps and components of the novel framework include the following:

1. Modeling: develop and train predictive models using historical data 2. Real-Time Prediction: apply the predictive models to process real-time data and anticipate potential problems in real-time 3. Real-Time Optimization: determine and execute the optimal actions needed to avoid the problems ant...