web start cache updates uses small % of avail network bandwidth

[Stephen Bannasch]

I used the http proxy tool Charles to examine the updating of jars from the jnlp server.

The data below were collected using my 3 Mbps DSL connection from home and running an otrunk-example which uses the all-otrunk.jnlp.

Summary: 23 jar resources were updated: size: 641k; time: 46s; rate 14.3 kBps

[Stephen Bannasch]

Here's the same test data comparing download speed to size of the response body.

Except for the outlier there is a very clear relationship between response-body size and download speed.

It would be worth investigating how much download speeds would benefit if we delivered a small number of large jars that combined the larger collections of jars we now distribute. This could be done using jarjar.

[Stephen Bannasch]

Here's the same test data but processed into a histogram showing the distribution of download speeds for all the jar and jardiff resources downloaded divided into 1k chunks

The first graph includes the otrunk.jar outlier, the second graph doesn't.

I suspect that there was a short network delay that caused the otrunk.jar outlier.

The area in each bar in the histogram is proportional to amount of data downloaded at that speed.

Even in the second graph you can see that an appreciable amount of the data is transferred at speeds less than 20 kBps.