That's great news. Thanks for the update, Stephen. Berkeley, are there any benchmarks for performance of WISE-3? I wonder how these performances improvements would affect those.
-t On Tue, Feb 24, 2009 at 12:50 PM, Stephen Bannasch < [email protected]> wrote: > > > Here's some info from recent "incremental improvements" we've made > for UDL that can work for any SAIL/OTrunk instance. > > The first two are applicable to any SAIL instance, the third is > specific to SAIL/OTrunk instances. > > 1) Local installer for Java Web Start Resources > > We try to have one frozen jnlp at the start of a semester for UDL and > this means that after the installer is run on a computer new jars > will not need to be downloaded (unless we've fixed a bug that needs > to be deployed). > > Here's a quick run-down on the Installer: > > The installer comprises a zipped archive of all the jars needed for > one or more JNLPs, and a small Java program that can start a Jetty > server and use Web Start to install the jars to the user's Web Start > cache. > > An example of this for the UDL project can be found at > http://udl.concord.org/install/ > > The installer for that project (probably the largest one) is about 80 > MB, and takes about 2-3 minutes to install to an empty Web Start > cache (shorter if there are already versions of the jars). > > As for installing to a folder outside of the user's home folder (so > that it can be installed once per computer), there is a page on that > here: > > http://confluence.concord.org/display/CSP/Deploying+Java+applications+onto+local+computers+in+schools > > With a frozen JNLP and the installer already run for a computer/user, > We estimate that it takes between 15-60 seconds for a student on a > slow network to start an activity, depending on how many users are > running and how much learner data they are downloading. > > 2) Fix continued activity slowdown during session. > > It turns out that the whole console log was being re-written EVERY > TIME a new line was written to the log. For projects that generate a > great deal of log output (true for the VLE) this obviously becomes > slower and slower as the log gets larger. > > See more details here: > > Running as the anonymous user from the DIY has long and increasing > delays. > https://jira.concord.org/browse/UDL-55 > > 3) Make second and subsequent startups faster (also makes within > session navigation faster). > > Add new JRes protocol for getting resources from jars > http://jira.concord.org/browse/ALL-61 > > This allows us to move external resources such as scripts and common > images into the signed jars stored in the users Java web start cache. > > Here are some tests we've run locally where the conditions are: > > # The local users Java web start cache is up-to-date > # The network speed from the CC servers to the users computer is > simulated as: 512 kbps, 50% utilization, 120ms network latency > > The limited network speed is simulated with the Java http network > proxy software called Charles. > > The idea is to simulate a student starting a UDL activity in a school > with a plausible but moderately-loaded network. > > Using the older codebase: > > UDL Fall 2008 Clouds 3-4: 15.5s > > Using the newer codebase and the same activity with scripts and > resources moved into jars to minimize initial downloading. > > UDL Spring 2009 Clouds 3-4: 11s > > Simulating a network speed of a 57kbps modem at 50% utilization the > startup time drops from 5 to 2 minutes. > > The overall speedups are actually more impressive because in the > older system scripts and many images would be reloaded over the > network as the student moved from page to page. The measurements > above don't describe those speedups. > > > > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "SAIL-Dev" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/SAIL-Dev?hl=en -~----------~----~----~----~------~----~------~--~---
