March 5, 2019 | Rodrigo
We’re excited to bring on our first customers to the preview release of our Workspace service on Azure. In the last few months we announced our partnership with Microsoft while working on the transition of AnalytiXagility as a whole to Azure and engaging with customers along the way.
This week, early adopter customers will get a chance to preview the service running on Azure. This allows customers to use a Workspace to explore:
- Integration with machine learning DevOps capabilities for automating model training and deployment
- Scaling out to high performance computing workloads using batch computing with on-demand hundreds of cores, terabytes of data
- Advanced data science using graphics processing units (GPU)
- Using Windows and Linux Virtual Desktops to access commercial and open source analytics software
- Replicating analysis previously undertaken on our existing implementations, prior to General Availability of AnalytiXagility on Azure
One of the first goals is to road test the preview release for usability and performance. In upgrading our platform to public cloud, we are taking the opportunity to re-engineer our systems to provide more on-demand resources transparently to the user. Users of our R console, Data Table Analytics and analytical database should notice faster start up times and execution for the these components.
Traditionally statisticians and data scientists have tended to work with files more than databases. As a result, the AnalytiXagility Workspace contains file storage, which in the cloud this can come in a number of flavours. At this preview release, Workspaces will include a traditional file system (for scripts and documents) as well as an object or ‘blob’ store. The latter can be more scalable and cost-effective as well as integrating with Azure cloud-based APIs such as machine learning to extract information from images or for bioinformatics on genomic data. However, existing software (e.g R studio) may not be adapted to interact with blob stores, so there’s still a need to provide traditional file systems. We’ll observe how users work their way around these differences and streamline the experience where we can.
Based on extensive customer demand we re-introduce Virtual Desktops which are fully integrated into the Workspace but with important improvements. We have previously provided access to Virtual Desktops using a remote desktop client. With this release, users can now connect to a Virtual Desktop through their browser and interact directly with software without installing any other software. Full screen mode is available too. We’ll offer Windows and Linux desktops on-demand as upgrade options for the Workspaces.
With the Virtual Desktops, we can start developing useful and interesting integrations with other services in the cloud. One of the first virtual desktop templates we will trial is the fully loaded Azure Data Science VM and we’re also looking at extending the Workspace to use the Machine Learning Service and DevOps for automating machine learning and the data science process for teams.
In the hospital context, our customers ultimately want to apply data science in the clinical workflow. Increasingly, hospitals will adopt SMART on FHIR as a way to customise patient record systems (EPRs) for specialty use cases. As part of our HDR UK sprint exemplar with Great Ormond Street Hospital we’ll be developing that integration by demonstrating integration with the Azure API for FHIR(r) and the App service.
As early adopters test the service, we expect to open the service up to pilot projects and by this summer to all customers. We look forward to updating this blog with progress towards that goal. Please get in touch if you have a project that fits the early access programme for the preview – we have some places left for interesting use cases!
At the same time, we continue to release updates to the main Workspace. There’s a parallel track on user experience improvement. The Workspaces 1.24 release was deployed last week and includes the first elements of a new look and feel.