1 person found this helpful
1) There are no real migration steps other than to update your Channel configurations to point to NIO endpoints versus the Servlet endpoints used in BlazeDS.
See the Channel configuration documentation. http://help.adobe.com/en_US/LiveCycleDataServicesES/3.1/Developing/WSc3ff6d0ea77859461172e 0811f00f7045b-7fe8Update.html
2) You are confusing LiveCycle Data Services (LCDS 3.1) with LiveCycle Enterprise (LC ES 2). Adobe does make it confusing because LCDS 3.1 is part of the LC ES2 suite.
lcds 3.1 is current the current version sold although that is less than obvious from the product page. http://www.adobe.com/cfusion/tdrc/index.cfm?product=livecycle_dataservices
Thanks for the quick response. I have a followup question. I'm not the Flex developer, so pardon my ignorance, but does the NIO connection info that you pointed me to also address setup of Stomp connections? I asked about setting up NIO connections in context of web client connections, but the actual area where one of our customers is hitting connection limits is Stomp server-to-server connections for publishing notification messages. Once in awhile the Stomp connections are not being released when we close the connection from our PHP app, and in those cases, the connections remain open forever on the Tomcat server unless/until Tomcat is shutdown and restarted, so eventually, over a period of several weeks, they run into the connection limit simply because the Blaze server is occasionally not shutting one down, and this builds up over time to several dozen stale connections that start getting in the way of new connections. That's the actual pain point that we need to address with this customer. They have enough streaming connections from the Flex client, but these stale Stomp connections wind up breaking our notification messages after awhile (I've got them restarting Tomcat/Blaze once a week to prevent the buildup). So I figured that Java NIO would help with this issue by providing a much higher connection limit, and maybe we'll get lucky and the buildup of stale connections won't even occur. And then, of course, we can support more streaming connections too. So after my long-winded explanation, can you confirm whether the article you pointed to also applies to Stomp connnections?