On Some Peculiarities of Dynamic Switch between Component Implementations in an Autonomic Computing System
Igor Mackarov (Maharishi University of Management)

TL;DR
This paper analyzes the behavior of a delta algorithm for autonomic switching between component implementations in client-server systems, highlighting specific conditions where it may perform suboptimally and proposing criteria for its success.
Contribution
It introduces a criterion for evaluating the success of the delta algorithm based on accumulated implementation costs, considering various system parameters.
Findings
The algorithm can behave unexpectedly under certain request cost and system parameter combinations.
A sufficient criterion for successful switching is proposed based on cost difference analysis.
Practical evaluation suggestions are provided for real-world system implementation.
Abstract
Behavior of the delta algorithm of autonomic switch between two component implementations is considered on several examples of a client-server systems involving, in particular, periodic change of intensities of requests for the component. It is shown that in the cases of some specific combinations of elementary requests costs, the number of clients in the system, the number of requests per unit of time, and the cost of switch between the implementations, the algorithm may reveal behavior that is rather far from the desired. A sufficient criterion of a success of the algorithm is proposed based on the analysis of the accumulated implementations costs difference as a function of time. Suggestions are pointed out of practical evaluation of the algorithm functioning regarding the observations made in this paper.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsDistributed and Parallel Computing Systems · Distributed systems and fault tolerance · Software System Performance and Reliability
