Completion Report
Scope
This project was to deliver a new service : a computational cloud for research. An OpenStack cloud service toprovide a scalable computational cloud platform accessible directly by researchers.
The project was to also increase the computational power of the Eddie 3 service via incorporation of new video cards and extended storage.
Objectives
Pre-Release Private Cloud Service available for research services consultants use by the 4th July and a full service in place for the 1st August. Achieved
An enhanced Eddie Service by end of October. Achieved (partially, completed February)
Deliverables
A new Cloud Service offering. Hadoop and Spark processing environments via the OpenStack cloud service. Delivered. New Cloud Service - Eleanor - launched on 14/2/17.
An R Studio deployment to the OpenStack cloud service. Delivered. ( Currently the R-Studio is a one off for MVM but could be made generally available).
An enhanced Eddie Service.
Upgraded memory per node available on the Eddie worker nodes to allow for larger virtual machine deployments within the cloud. Increased working memory available within the Eddie Service. Delivered.
Extended the data processing capability of Eddie with the expansion of the available GPGPU processing cluster. Delivered. (with some additional deployments to complete the extension).
Extend the data processing capability of Eddie by adding two large 3TB memory systems. Delivered.
Benefits
Having a computational cloud gives flexible access to computational resource Access to new data science analysis tools
Extend Eddie 3 Ability to work with larger data sets Production scale run GPGPU analysis with particular focus on Machine Learning
These benefits and their management and realisation will be available on the ITI Sharepoint space.
Success Criteria
Researchers in the Data Science Community can create customised environments offering computational resource in the cloud. Success.
Researchers can run GPGPU analysis and Machine Learning. Success.
Lessons Learned