Blogs & News
Aridhia is Introducing Containers to Workspaces
Historically, we have allowed users to bring their own compute to their secure workspace environment. So far, we have offered this functionality though Virtual Machines (VMs) which give good flexibility and allow users to run their apps or software in a collated environment against their workspace data. We have improved the experience over time, moving from remote desktops to embedded VMs in the workspace, but we think it might be time for the next improvement: How can we make it easy and quick for users to bring their own tools and apps to their own secure data and also share these tools with others?
Health(care) data often demands specialist analysis in order to gain useful insight. Data can be sensitive, large, complex and require expert insight in order to understand it and we believe that we are set up to deal with this data using our state-of-the-art Digital Research Environment (DRE). However, sharing insights into the data and reproducing experiments and analysis is an ever-expanding issue to solve; it is not impossible but it could certainly be made easier.
Data analysts and scientists have increasingly been developing apps which can be used to gain insights into their data. These apps can be rather large and complex but as long as people are working with data in the right format, this may be the best way for them to understand the results of their work. The setup for a particular analysis project can take months and may involve a long list of libraries and packages which need to be curated and maintained over the lifetime of a project. If you leave your work for even a short while before returning, you cannot be assured that your analysis will still run.
Additionally, researchers often want to share their analyses or techniques with contributors or colleagues who work in remote locations, often using different platforms or with a different setup. One way to share your own analytics is to containerise your program in order to ship it to others.
Containers allow a user to snapshot and package their setup at a particular moment. The container essentially works as a lightweight virtual machine which comes pre-configured with everything you need to do a particular job; all you need is a platform to run that container and you can then use new data to run the contained analysis.
They can also act as a vehicle to improve or ensure reproducibility of experiments; store your setup (including data if needed) at a particular point in time and all you need to do to run the exact same experiment again is plug in your container and let it run against your new data. Containers can be stored, shared and developed throughout the lifetime of the project and further.
At Aridhia, we are currently working on providing an environment within our Workspaces which allows users to bring a Docker container, plug it into the workspace and use it against the data stored there.
We see many advantages to providing this new functionality in our secure Workspaces. Within an organisation, creating a repository of these containers and allowing users to pull them into their workspace allows the secure sharing of knowledge and analysis as well as allowing the organisation to be reassured that their data is secure and safe. The repository could of course be curated to allow only approved or reviewed containers which pass security or other standards. For research groups, the ability to quickly set up and run or re-run an experiment is really exciting. Bringing that experiment to their own data in a secure environment, even more so.
Of course, we must make considerations: how do we ensure that workspace data is protected? How do we let organisations control which apps and containers can be deployed? How do we make it easy to develop and share these apps in our own ecosystem?
Having already run a (successful) pilot project internally, we are now seeking to roll out this functionality whilst maintaining security and traceability of work as well as workspace integrity.
More information about our current Roadmap, and this project in particular, can be gained through contacting us or commenting below. We would welcome your thoughts and insights.