The automagic part is a bit spooky when it comes to data shared among threads. If you have large reporting systems that share big data caches in the jvm, but that can process in multiple threads, doesn't this data have to get copied between all nodes needing it over the network? Seems like this could be a good solution for applications that are compute but not data intensive, and that don't require many shared resources.
We have one the biggest, most complicated java based web delivered applications on the internet, but our solution is to remove common state from the application processing machines-- they are stateless number crunchers. State (common data) is maintained in nodular cluster databases. Application load is easily distributed using load balancing in front of the application servers based on parameters like cpu and other resource load ( we use mod_athena from http://ath.sourceforge.net/ ). You can farm out to commodity systems this way.
Mark Wolgemuth, Production Architect Employease, Inc. ( www.employease.com )
i've set up a small farm of linux boxes for serving http, dns, and smtp/pop for employease. i've been using the HP LPr machines: 2U form factor, twin cpu configuration. they have a serial/console out that you can configure linux to use. we are going to send all console interfaces to a cyclades console management system. i'm not sure the LPr's support remote power cycle signals via this interface, but I know the Dell Poweredge 6350's we're using do. unfortunately, these machines currently run NT since Netscape Enterprise server is not yet ported to linux. however, we have it from an insider that Enterprise server will be available in the fall. in the meantime, we are going to test out the quad-xeon-cpu 6350's as linux servers, and will find out if its console interface will actually work with linux.
PS: Neither of these boxes care if they are headless, handless, or tailless.
Slashdot Mirror
The automagic part is a bit spooky when it comes to data shared among threads. If you have large reporting systems that share big data caches in the jvm, but that can process in multiple threads, doesn't this data have to get copied between all nodes needing it over the network? Seems like this could be a good solution for applications that are compute but not data intensive, and that don't require many shared resources.
We have one the biggest, most complicated java based web delivered applications on the internet, but our solution is to remove common state from the application processing machines-- they are stateless number crunchers. State (common data) is maintained in nodular cluster databases. Application load is easily distributed using load balancing in front of the application servers based on parameters like cpu and other resource load ( we use mod_athena from http://ath.sourceforge.net/ ). You can farm out to commodity systems this way.
Mark Wolgemuth, Production Architect
Employease, Inc. ( www.employease.com )
i've set up a small farm of linux boxes for serving http, dns, and smtp/pop for employease. i've been using the HP LPr machines: 2U form factor, twin cpu configuration. they have a serial/console out that you can configure linux to use. we are going to send all console interfaces to a cyclades console management system. i'm not sure the LPr's support remote power cycle signals via this interface, but I know the Dell Poweredge 6350's we're using do. unfortunately, these machines currently run NT since Netscape Enterprise server is not yet ported to linux. however, we have it from an insider that Enterprise server will be available in the fall. in the meantime, we are going to test out the quad-xeon-cpu 6350's as linux servers, and will find out if its console interface will actually work with linux.
PS: Neither of these boxes care if they are headless, handless, or tailless.