Quantifying the impact of network latency on the end-to-end response time of distributed applications
First Claim
1. A method in a distributed system where relationship between a property of a first component of said system and an end-to-end metric of said system is determined by injecting a failure in said first component and identifying effect of said injecting on said metric, the improvement comprisingreplacing said injecting of a failure with injecting of a change in said component that is allows said distributed system to continue operating at normal mode;
- andquantitatively evaluating said relationship.
1 Assignment
0 Petitions
Accused Products
Abstract
A method for measuring system response sensitivity, using live traffic and an analysis that converts randomly arriving stimuli and reactions to the stimuli to mean measures over chosen intervals, thereby creating periodically occurring samples that are processed. The system is perturbed in a chosen location of the system in a manner that is periodic with frequency p, and the system'"'"'s response to arriving stimuli is measured at frequency p. The perturbation, illustratively, is with a square wave pattern.
-
Citations
34 Claims
-
1. A method in a distributed system where relationship between a property of a first component of said system and an end-to-end metric of said system is determined by injecting a failure in said first component and identifying effect of said injecting on said metric, the improvement comprising
replacing said injecting of a failure with injecting of a change in said component that is allows said distributed system to continue operating at normal mode; - and
quantitatively evaluating said relationship. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20)
- and
-
21. A method for measuring responsiveness of a system that receives requests at non-regular intervals, comprising the steps of:
-
undertaking a training time interval during which (a) intervals between arriving requests are noted, and (b) a statistical analysis is performed to identify an interval Δ
Ts within which there is a selected probability in excess of a chosen value that at least one request will arrive; andmeasuring an attribute that represents average responsiveness of said system to requests arriving during each Δ
Ts interval. - View Dependent Claims (22, 23, 24, 25, 26, 27, 29, 30, 31, 32, 33, 34)
-
-
28. The method of 26 where an attribute is evaluated for a frequency that is related to kd.
Specification