Hi,
I am planned to use the customized socket factory for the secured RMI communication by encryption of the data communicated. Is it an advisable to go with that.
if it is an advisable one, Please tell the valid encryption algorithms for that process.
Please forward the benchmarking of RMI with customized socket factory communication
it's very urgent.
Thanks in advance
Regards,
RedRose
> I am planned to use the customized socket factory for
> the secured RMI communication by encryption of the
> data communicated. Is it an advisable to go with
> that.
It's the only way.
See http://java.sun.com/j2se/1.5.0/docs/api/javax/rmi/ssl/package-frame.html
> if it is an advisable one, Please tell the valid
> encryption algorithms for that process.
See http://www.ietf.org/rfc/rfc2246.txt.
> Please forward the benchmarking of RMI with
> customized socket factory communication
Come off it. This is a volunteers' forum. How about you find some yourself?
> > I am planned to use the customized socket factory
> for
> > the secured RMI communication by encryption of the
> > data communicated. Is it an advisable to go with
> > that.
>
> It's the only way.
But i heard that, if we used customized socket factory classes we have to lose some of the default properties of RMI.
eg., HTTP tunneling won't be allowed
>
>
> > Please forward the benchmarking of RMI with
> > customized socket factory communication
>
> Come off it. This is a volunteers' forum. How about
> you find some yourself?
I am in need to optimize the RMI performance, can i do the "Remote Stub Caching" process to reduce the stub invocation. If it's an advisable one please send some of code samples.
> But i heard that, if we used customized socket
> factory classes we have to lose some of the default
> properties of RMI.
> eg., HTTP tunneling won't be allowed
Yes. The RMI socket factory mechanism is violently overloaded to accomplish everything from SSL to firewall penetration. It can be made to support a delegation model so you can get some of the default behaviour back if really necessary. However if you're contemplating RMI over HTTP over SSL you can't do it anyway because AFAIK the default socket factory can't be made to agree with the rmiservlethandler or rmi-cgi script about when SSL should be applied. I am very hazy on these details as it is many years since I looked at this but that's my recollection. It's in my book and nobody has challenged it ;-)
> I am in need to optimize the RMI performance, can i
> do the "Remote Stub Caching" process to reduce the
> stub invocation.
It won't reduce the stub invocation [time] [?], it just reduces RMI Registry lookups. Not doing any more lookups than you need to is always a good idea. However you do have to do some: the first one, and a reacquisition lookup if you incur a NoSuchObjectException (meaning the stub is stale).
