Software Solution must determine the:
• number of nodes to complete the requested job either by count or translated by an established price per node; or, alternately as a part of the UI,
• time requirements for the job in order to determine number of nodes
for each job request.
Upon setting up of a session, software must then:
• identify the software needed from persistent storage to run on compute units pursuant to the job request;
• identify the assets for the job in already stored in object storage (part of the requirements for the job request along with payment information);
• identify how the processing are to be divided for processing; and,
• write and convey the job description – including the file paths of the assets and the software requirements and the designated number of nodes to be used for the job – to the hardware master node for processing.
Following processing, software must then:
• assemble processed segments into final output;
• return final output to object storage for further disposition – whether downloading to user requesting the job or further processing.
Software must also be able to connect to PaaS process that may procedurally precede the subject PaaS as well as that might follow the subject PaaS running on the same IBM Bluemix infrastructure, i.e., software must also be able to plug into other processes with common data.
All code to be delivered via GitHub upon conclusion after full dev, testing and QA