1. A Framework for Automatic Resource and Accuracy Management in A Cloud Environment
Smita Vijayakumar

2. Outline Background Research Goals CPU Resource Allocation
Accuracy Management
Experimental Evaluation
Conclusion

3. Outline Research Goals Accuracy Management Experimental Evaluation
Background
Data Streams
Virtualization
Dynamic Resource Allocation
Accuracy Adaptation
Research Goals
CPU Resource Allocation
Accuracy Management
Experimental Evaluation
Conclusion

4. Data Streams Sequence of data packets in transmission
Time interval between successive packets never a perfect constant
This leads to Jitter 4

5. Examples of Data Streams
Live Camera Captures
Telecommunication Networks
Stock Markets
Network Monitoring
Video Streaming Applications
Business Transaction Flows 5

6. Properties of Data Streams
Require Real-Time Analysis
Time-Varying Characteristics
Time-Varying Resource Requirements 6

7. Resource Allocation: Guiding Principles
Utility Computing
Pay-as-you-go
Two desirable features :
Automatic Resource Allocation
Eliminates User intervention
Dynamic Resource Allocation
Better Resource Management
Lesser Overall Costs

8. Virtualization
Software implementation of a machine that executes programs
Resources can be allocated in a fine-grained way
CPU cycles can be changed

9. Why Dynamic Resource Provisioning for Data Streams
Sufficient Data to Process -> Max CPU Utilization
Insufficient Data -> Wasted CPU Cycles and Unwanted Cost
Varying Data Rates and Characteristics 9

10. CPU Allocation Algorithm
Near-Optimal resource allocation at bottleneck stage by constant monitoring of processing load
Design Principle: Match processing speed with data arrival rate

11. Accuracy Adaptation Consider Adaptive Streaming Applications
Such application have Adaptive Parameters that determine accuracy of the application
Parameter values can be set according to user-desired accuracy levels and resource budgets

12. Example of Adaptive Application
Random stream of integers
Application retains the average of every third integer in the stream
Adaptive Parameter, sample= 1/3
If higher accuracy is desired, sample can be set to ½ or 1
But then, that requires more CPU resources

13. Accuracy in Data Stream Processing
Accuracy-specific processing
Requirement: User-defined processing accuracy should be met
Require corresponding Resource Allocations
Final cost determined by amount of resources allocated

14. Calculating Current Application Accuracy
Application developer provides Accuracy Function
Many methods of calculating accuracy:
Method of direct comparison with input data
Not always viable
Method of correlation with more fine-grained processing
Process data with current adaptive parameters
Process same data set with adaptive parameter set to greater accuracy
Compare results

15. Example of Accuracy in Adaptive Application
Process batch with current value sample =1/3
For same data set
Set sample = 1and find new average
Accuracy = f(avg, higher_avg)
If Accuracy < Accuracy Goal, set sample = 1/2
Repeat adapting sample

16. Research Goals
Framework for providing Accuracy and Resource Management in Cloud Environment
Accuracy Management
Convergence to application-specific accuracy goal
Maintain user-specified accuracy requirement for the entire duration of run
CPU Management
Converge to near -optimal resource allocation by constant monitoring of load characteristics

17. Outline Background Research Goals CPU Resource Allocation
Accuracy Management
Experimental Evaluation
Conclusion

18. CPU Allocation Algorithm
Monitor current load statistics
Buffer Write Time
Processing Time
Time-Averaged rates
Average data rate over a time window
Update CPU allocation
Time- Averaged pattern indicates decrease or increase in data flow
Continuous Monitoring and Action
Arrive at most optimal CPU Allocation

19. CPU Allocation Algorithm
Resource Allocation Adjustments:
Coarse Multiplicative Increase
Fine Linear Increase
Fine Linear Decrease
Coarse Linear Decrease
Inspired by TCP Congestion Control

20. CPU Allocation Algorithm
Met Accuracy Goal?
Sleep and awaken periodically
Adjust CPU Allocation
Met Allocation Needs?
Yes No
Yes No

21. Outline Background Research Goals CPU Resource Allocation
Accuracy Management
Experimental Evaluation
Conclusion

22. Accuracy Management Checks periodically for accuracy level
Re-computes application accuracy
If less than specified value then
Adjust adaptive parameters
Repeat
Once target accuracy is achieved, wakes up after every 500 rounds of processing

23. Accuracy Adaptation: Design
Get Current Application Accuracy
Met Accuracy Goal?
Sleep and awaken periodically
Yes No
Adjust Adaptive Parameters

24. Interaction between Components
Process Data Block
If baseline accuracy not met
Accuracy Module adapts till accuracy is met
State: Accuracy Met
Else, periodically monitor accuracy
Periodically CPU Manager wakes up
Checks if accuracy goal is met
Checks CPU resource allocation

25. Outline Research Objectives Introduction to Cost Framework
CPU Resource Allocation
Accuracy Management
Experimental Evaluation
Static Experiments
Dynamic Experiments
Conclusion

26. Experimental Focus Static Experiments
Constant Data Rate And Characteristics
Dynamic Experiments
Varying Data Rates and Characteristics
Evaluate
Accuracy Adaptation for User-Specified Accuracy
Converge to near-optimal CPU Allocation

27. Streaming Applications
Multi-staged pipelined processing
Two streaming applications considered:
CluStream
Intermediate Microclustering of data
Approx-Freq-Counts
Mining most frequently seen itemset within permissible error

28. Experimental Setup Virtualization Technology: Xen
Ideal CPU Usage: Xentop
Applications initialized to values corresponding to least accuracy
Communication between management node and processing nodes using UDP

29. CPU Convergence Experiments
Experimental Results
CPU Convergence Experiments

30. CluStream Static Convergence
36.0%
35.6%
Convergence to an ideal 36% CPU Allocation

31. Approx-Freq-Counts Static Convergence
71.3%
71.2%
Convergence to an ideal 71% CPU Allocation

32. CluStream Dynamic Convergence
Convergence to CPU Allocation

33. Approx-Freq-Counts Dynamic Convergence
Convergence to CPU Allocation

34. Accuracy Convergence Experiments
Experimental Results
Accuracy Convergence Experiments

35. CluStream Static Accuracy Adaptation
Accuracy Adaptation for 1.2MBps and 6MBps data rates

36. CluStream Static Accuracy Adaptation
Accuracy and CPU Allocation Adaptation for 1.2MBps and 6MBps data rates

37. Approx-Freq-Counts Dynamic Accuracy Adaptation
Spread Distb
Sharp Distb
Spread Distb

38. Approx-Freq-Counts Dynamic Accuracy Adaptation
Sharp Distb
Slow Data Rate
Spread Distb
Fast Data Rate
Sharp Distb
Slow Data Rate

39. Conclusion
A framework for automatically and dynamically managing resource allocations on cloud environments
Eliminates manual intervention
Ensures user-specified accuracy maintained
Converges to near-optimal resource allocation
Adapts to varying data stream characteristics
Low Overheads: Within 2% ideal resource allocation

40. Thank You!