Visually Compare Retry Algorithms
compareretries.com
Visually Compare Retry Algorithms
1–6 of 6 posts
Re: Visually Compare Retry Algorithms
#2This is awesome. It's nice to have a tool to show people why they shouldn't have non-capped exponential retries.
It would be even nicer if adding configurations updated the URL so you could actually share the configuration with others. Oh, and a log(time) axis for better visualizations of non-linear retries.
For now, just set a 5x exponential retry of 100ms...
Re: Visually Compare Retry Algorithms
#3This is awesome. It's nice to have a tool to show people why they shouldn't have non-capped exponential retries. It would be even nicer if adding configurations updated the URL so you could actually share the configuration with others. Oh, and a log(time) axis for better visualizations of non-linear retries. For now, just set a 5x exponential retry of 100ms...
It doesn’t seem so bad if you set the factor to something more reasonable like 2?
Re: Visually Compare Retry Algorithms
#4+1 to the other commenter. Please, make the URLs shareable. Maybe use the PlantUML's technique as the inspiration.
Re: Visually Compare Retry Algorithms
#5This is awesome. It's nice to have a tool to show people why they shouldn't have non-capped exponential retries. It would be even nicer if adding configurations updated the URL so you could actually share the configuration with others. Oh, and a log(time) axis for better visualizations of non-linear retries. For now, just set a 5x exponential retry of 100ms...
It doesn’t seem so bad if you set the factor to something more reasonable like 2?
Why would you retry after 2ms? In most cases, the odds are extremely likely you’d have the exact same outcome; but it depends on your application.
Re: Visually Compare Retry Algorithms
#6Earlier quoted context omitted.
It doesn’t seem so bad if you set the factor to something more reasonable like 2?
Why would you retry after 2ms? In most cases, the odds are extremely likely you’d have the exact same outcome; but it depends on your application.
2 is the factor, not the initial delay.