SOSCIP’s NDRIO: Engage DRI White Paper

SOSCIP Consortium on the Importance of Driving Industry Innovation with Canada’s New Digital Research Infrastructure

SOSCIP recently submitted a White Paper on the future of Digital Research Infrastructure to NDRIO as part of their Needs Assessment initiative


The following is an excerpt of the white paper:

Accessing DRI Tools, Services and Resources

Access to DRI tools, services and resources needed for the industry-academic research we support typically depends upon the area of research. For example, large companies who have project work in aerodynamics or materials science usually need access to an HPC-type machine but do not have large enough capacity for their growing demand. On the other hand, SMEs who are in an AI field generally do not have access to high-end GPUs. Commercial AI R&D software/services like Amazon Sagemaker and Google Datalab are not available on the current infrastructure and would be too expensive for SMEs to use intensively on commercial cloud. This appears to be a growing need that cannot be currently met.

Challenges to accessing and using DRI

There are a number of significant challenges that currently exist using the current DRI tools. One challenge is that of out-of-service situations typically due to network, power and cooling that can occur and sometimes can take a few days to get resolved. For this reason, the current infrastructure is not reliable enough for a company to run any 24/7 client-facing service such as web portal, public database, real-time request or analysis. Technical support is generally not available other than working hours which is usually business hours on weekdays, and depending upon the client’s time zone, support may not be available.
Another significant challenge for SME clients is that they often cannot afford highly-skilled development personnel to smoothly migrate their work from a research-oriented infrastructure to a publicly available infrastructure without proper training and hands-on support. Obtaining sufficient allocation on existing infrastructure is another challenge for academic-company collaborations for industry-driven/commercially driven projects. Recently, companies have begun more often seeking and specifically asking for secured cloud-based infrastructure. Research-oriented HPC-type machines appear to not be secure enough for them.

Read the Full White Paper Here