When the World is Not Enough. Enabling the Science then Somehow Dealing With It

Something happens when we enable great things through the judicious use of technology in the modern scientific era. Quickly, we find ourselves with problems that were not problems before, trying to answer questions that the previous platforms and enabling technologies could not answer. Human nature and the mind of every scientist dictates that we will push things that little bit further, or prod something in a slightly different direction, even if we know it might break things.
Whether we like it or not, as systems administrators, CIO’s and technology leaders, this inevitably creates more and more significant demands upon the platforms we’ve created for our users. This presentation will detail and articulate what happens when you offer people the world in scientific research tools and platforms, and they subsequently take it.
The Queensland Brain Institute has changed infrastructure several times in an attempt to keep up with the science that takes place within the walls. Valuable lessons have been uncovered and learned as a consequence. Managing scientific infrastructure is a science in itself. Looking after the specialised networks, storage and computing infrastructure that power the research has its own set of eccentricities. Truly ubiquitous workflows, communications and interactions don’t occur easily, nor do they work as seamlessly as product marketing would have us believe, when faced with particularly large-scale propositions.
Through robust description and analysis of environmental patterns, heuristics and analytics, we hope to present a view of research that demonstrates just how easily confounded generic enterprise growth pattern calculators can be, as well as the fragile nature of what the vendors think they know about providing appropriate silicon to power the needs of research in the big-data/big-switching and big-compute era. The audience can expect to see a demonstration of this in action as well as some critical analysis of how it impacts upon resources.
If there is no easy way to predict growth in such extreme and volatile environments, how do we scale infrastructure appropriately? If indeed it’s that hard to contain the needs of research, then why are so many large centrally/government funded projects suggesting they have the answers? Why are suggestions being made about silver-bullets that solve the big data collaboration and ubiquitous/seamless interaction problems?
We will attempt to answer some of these bigger questions about the management, expectations and the reality of such situations, through experience and research.