[CASSANDRA-2597] inconsistent implementation of 'cumulative distribution function' for Exponential Distribution

• Type: Bug
• Status: Resolved
• Priority: Low
• Resolution: Fixed
• Fix Version/s: 0.8.1
• Severity: Low
• Assignee: paul cannon
• Reporter: Jonathan Ellis

Authors

paul cannon

Reviewers

Brandon Williams

• Created: 03/May/11 19:28
• Updated: 16/Apr/19 09:33
• Resolved: 23/May/11 19:44

Description

As reported on the mailing list (http://mail-archives.apache.org/mod_mbox/cassandra-dev/201104.mbox/%3CAANLkTimdMSLE8-z0x+0kvzqp7za3AEMLaOFXvd4Z=[e-mail]%3E),

I just found there are two implementations of 'cumulative distribution
function' for Exponential Distribution and there are inconsistent :

FailureDetector

org.apache.cassandra.gms.ArrivalWindow.p(double)
   double p(double t)
   {
       double mean = mean();
       double exponent = (-1)*(t)/mean;
       return *Math.pow(Math.E, exponent)*;
   }

DynamicEndpointSnitch

org.apache.cassandra.locator.AdaptiveLatencyTracker.p(double)
   double p(double t)
   {
       double mean = mean();
       double exponent = (-1) * (t) / mean;
       return *1 - Math.pow( Math.E, exponent);*
   }

According to the Exponential Distribution cumulative distribution function
definition<http://en.wikipedia.org/wiki/Exponential_distribution#Cumulative_distribution_function>,
the later one is correct

... however FailureDetector has been working as advertised for some time now. Does this mean the Snitch version is actually wrong?