Print

[yin]

Hi,

I'm conducting a feasibility study on application of JPPF in a p2p inventivised volunteer grid computing context

1) Is there a credit mechanism similar to BOINC credits that could be used for incentives?
2) Does anyone have documented experience with p2p JPPF volunteer grid and if so, how does it perform?

Thanks
 

[lolo]

Hello,

1) Is there a credit mechanism similar to BOINC credits that could be used for incentives?

Unfortunately, there is no such built-in mechanism in JPPF. It would have to be implemented in an ad-hoc fashion, which should be relatively easy given the flexibility of the JPPF APIs.
For instance: since a JPPF volunteer computing node is basically a standard JPPF node with a specialized configuration, it is easy to configure each node with a user/team name which would receive the credits. Each task (or unit of work) can then easily fetch the name and return it as part of its results. The client would then be able to process it to effectively compute the credits.

In particular, JPPF does not measure the capabilities of a node prior to sending work, like BOINC does by running benchmarks to measure the performance of a host. There is a load-balancing mechanism, which can adapt to the past performance of each node, but obviously that's a very different design from Boinc.

I hope this clarifies.

Sincerely,
-Laurent

[yin]

Hello,

1) Is there a credit mechanism similar to BOINC credits that could be used for incentives?

Unfortunately, there is no such built-in mechanism in JPPF. It would have to be implemented in an ad-hoc fashion, which should be relatively easy given the flexibility of the JPPF APIs.
For instance: since a JPPF volunteer computing node is basically a standard JPPF node with a specialized configuration, it is easy to configure each node with a user/team name which would receive the credits. Each task (or unit of work) can then easily fetch the name and return it as part of its results. The client would then be able to process it to effectively compute the credits.

In particular, JPPF does not measure the capabilities of a node prior to sending work, like BOINC does by running benchmarks to measure the performance of a host. There is a load-balancing mechanism, which can adapt to the past performance of each node, but obviously that's a very different design from Boinc.

I hope this clarifies.

Sincerely,
-Laurent

Hello Laurent,

Thank you for the prompt and comprehensive response.

Keep up the brilliant work!