Hello KP,
May I assume that you already had a look at our sample
large portfolio update simulation?
If not, I invite you to do so, I believe it matches the kind of application you describe for your project.
To answer your questions:
1) Would the potential network activity between driver & nodes impact performance of my application?
Yes, it definitely would. The impact would be most significant in the case your jobs have very few short-lived (e.g. in the milliseconds range) tasks, and least significant when the jobs have many long-lived tasks.
Some of the main factors impacting the network overhead would be: the network quality/speed, the tasks footprint, the load-balancing algorithm used, how you chose to distribute the tasks among the jobs (for instance you may chose to "batch" multiple tick updates together, based on a number of ticks or a time interval).
2) Do you have any case studies of JPPF used in real time applications? Could you please share them with me?
3) Do you have any case studies of JPPF used in an OMS application? Could you please share them with me?
Unfortunately, there is no such case study that I can share with you. Sorry for the inconvenience
4) Could you please recommend any other framework / grid toolkit that could be of help in my project?
Some of the frameworks in this area:
- GridGain, Apache Hadoop (open source, based on map/reduce model)
- GigaSpaces (commercial, based on Jini/JavaSpaces)
- Hazelcast (open source, distributed Java collections and executor services)
I hope this helps.
Sincerely,
-Laurent
