Re: Scaling Reggie

Hi, Dennis; there can be an issue when direct ByteBuffers are allocated
and not reused (as, if I recall correctly, the JERI NIO impl does) because with direct
ByteBuffers there is no correlation between the amount of memory the ByteBuffer
handle occupies on the Java heap and the amount of memory that has actually been
allocated on the C heap.  The memory on the C heap does not get deallocated until
the ByteBuffer handle gets garbage collected, but the garbage collector might not do
a collection because there is enough memory on the Java heap.  There are some
details at:

http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=4857305

----- Original Message -----
From: Dennis Reedy <[email protected]>
Date: Tuesday, September 18, 2007 3:46 pm
Subject: Re: Scaling Reggie
To: [email protected]


> >>
>  >
>  > Reggie per se does not make a copy of the bits that comprise the
>  > serialized
>  > service proxy; not sure if the JERI dispatch layer might be doing
>  > some copying
>  > of the service proxy bits when dispatching concurrent requests;
>  > others in the
>  > list can probably address better.  Is NIO being used by any chance?
>
>  Hi Juan,
>
>  Interesting you mention this, the answer is yes. Why would NIO play
>  into this?
>
>  Thanks
>
>  Dennis
>
>  --------------------------------------------------------------------------
>  Getting Started:     http://www.jini.org/wiki/Category:Getting_Started
>  Community Web Site:  http://jini.org
>  jini-users Archive:  http://archives.java.sun.com/archives/jini-users.html
>  Unsubscribing:       email "signoff JINI-USERS"  to [email protected]
>

--------------------------------------------------------------------------
Getting Started:     http://www.jini.org/wiki/Category:Getting_Started
Community Web Site:  http://jini.org
jini-users Archive:  http://archives.java.sun.com/archives/jini-users.html
Unsubscribing:       email "signoff JINI-USERS"  to [email protected]