Eucalyptus is running

Finally we have managed to make our Eucalyptus private cloud running. The first matlab scripts processed thousands of pictures. It took us quite some time to make this happend. We had many problems with network and routers configuration, with NFS, disk overflows etc. All typical network computing problems, most of them solved by now. But we are happy we are producing useful results on computer which were so fare idling.
The current cloud setup is used for number crunching only, for image processing. Each experiment is described by a matlab script.  There is currently available only one machine image, which includes installed matlab. To start and control execution we have developed a SaaS application uploading and managing the experiments. It also works with AWS. We are running the application on the same box as the cloud server. It is written in R&R. The UI is designed for inexperienced users. The application messaging is build on top of the RabbitMQ. The input and output files are read and written to and from users subdirectories over NFS.
Further development is still needed to make the cloud much more efficient. We need to include some monitoring to control the usage of the cloud. We need to optimize memory, networking and speed of execution. We need to control user access etc.
We have successfully proved, that we have chosen the right solution. Our first test user managed to use the system almost immediately and produced useful results. The next step is to add another two user with no experience and collect all complaints, wishes, bugs etc. to continue in the development of the controlling application and make the whole system stable and robust.