Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : All About Sun Microsystems -- Ignore unavailable to you. Want to Upgrade?

To: Stormweaver who wrote (21714)	10/24/1999 8:54:00 PM
From: paul	Respond to of 64865

".. Ideally I can plug an NT box into my node group of Sun machines and scale my distributed app.." "Ideally" this would be great. Microsoft announced Wolfpack (MSCS) back in '96 and all its capable of doing is failing one node over to another node - even Novell has been able to do this for years. You dont get any applications scalability with MSCS - the benefit of your front end is simply in using cheap PC's and sacrificing managability, consistent OS' from the front to back end and superior RAS and performance of Sun boxes. Distributed processing is not something anyone is against - nobody is saying everyone should run their application on a single E10K. Sun makes a lot of 1-4 CPU systems which are the default systems of service providers and ISP's and act as excellent front ends and application servers. You can add capacity with another clustered node all running a single image of Solaris - plus they can consolidate many little PC servers solving the software and logistical problem of server creep. A larger system such as an E10K can be used as the back end server which functions in a more predicatable environment - the benefit of the back end server is in its ability to scale seamlesly. Microsoft doesnt have a solution for the back end either, so at least their consistent.

To: Stormweaver who wrote (21714)	10/25/1999 4:07:00 PM
From: Robert	Read Replies (1) \| Respond to of 64865

Writing distributed solutions is tricky. I have been learning towards distributing program logic as well as a much simpler way to manage the program. This leads almost naturally towards software agents. Comparing SAP and programming work flow schedules and "intelligent" work agents able to route themselves, the latter makes more sense from several viewpoints. Intelligent agents would identify for themselves the next available server, reprogramming the agent would not require bringing down the server, new agents could be created at runtime, agents can negotiate between themselves precedence for access to resources, and agent behaviour would follow natural logic more closely as to what the human would do as an individual. I think hardware issues are really the least of the problems with distributed applications. Getting the architecture right is hard enough without worrying about anything below TCP/IP. The irony is that Java commoditises everything under it, so people should worry about what goes on top as value added.