How the cloud is making R&D easier

 

Industries underpinned by scientific processes – from pharmaceuticals to oil refining – have seen huge efficiency improvements from the digital revolution. Labour intensive processes once requiring endless experimentation can now largely be done in silico, with much greater efficiency, whilst also opening up new avenues for research beyond the capabilities of manual labour.

Sophisticated applications now do a range of bespoke data analysis tasks to help researchers make new discoveries: from experiment design and optimisation for chemical formulation analysis at one end of a product lifecycle to safety analysis for chemical manufacturing at the other.

These apps have mostly been set up to run on physical IT infrastructure. Many industries, life sciences and Oil & Gas amongst them, are currently undergoing (or soon to undergo) major overhauls of their IT systems, as old technology reaches the end of its useful life or becomes too slow for their advancing data analytics needs. One of the most important decisions will be how much to replace physical kit with hosted cloud services.

This major upgrade is a unique opportunity to get much more out of scientific applications by carefully moving them to the cloud, and so ensuring they deliver their full potential to the business, now and in future.

Many companies have been hesitant to move such apps to the cloud for a number of reasons. Some are concerned that moving such complex apps to new systems risks upsetting their configurations, others are understandably nervous about allowing business critical data off their servers, and some are simply worried about cost or resistant to change. However, ageing physical IT infrastructure limits these apps’ business potential. As companies reconsider how their IT can be better designed for greater business impact, a well-managed migration of scientific apps to the cloud should form an important part of this strategy.

Better in the cloud

Cloud is a much more natural home for these types of applications. Firstly, cloud gives unlimited and scalable computing power. As apps process ever more data, they become more and more power hungry, so long term scalability is important to manage progress. Of more immediate concern is that such apps often need to run batch processes over a few days. With physical infrastructure, this means either compromising capacity elsewhere or building in capacity you don’t need most of the time. In the cloud, you simply spin up some dedicated servers when you need them.

Secondly, cloud makes engagement easier. Many companies are keen to spread the use of these apps throughout their organisation, speeding research by allowing researchers access to the best tools. Gradual roll out is preferred as it will reveal who can benefit and who can’t and allow people to move at their own pace, rather than forcing them on everyone at once, regardless of need and readiness. Cloud allows you to easily scale up as your user base grows, without needing to do detailed assessments for years ahead.

Finally, it allows you to experiment with new approaches. Moving things to new infrastructure doesn’t always work as well as expected. This is a problem if you’ve spent a lot on that infrastructure. In the cloud, you can try an approach, and if it doesn’t work, you can pull it down and start again. With cloud, you can make sweeping ‘infrastructure’ changes, at push of a few buttons.

Moving apps to the cloud – the challenges

Although the advantages are numerous, the concerns companies have are real, and must be dealt with. Once the business decision makers have bought into the benefits, as outlined above, there are various practical hurdles.

As legacy IT infrastructure needs updating, Tessella has recently found itself helping several organisations, principally in life sciences, to navigate the process of moving apps to a publicly hosted cloud, and our experience has taught us a few approaches and shortcuts that can make this process much more straightforward.

Moving complex scientific processes to the cloud often involves a choice between Infrastructure as a Service (IaaS) and Platform as a Service (PaaS). IaaS can allow more direct mirroring of existing infrastructure, but will incur greater ongoing maintenance costs. PaaS provides the entire underlying infrastructure allowing you to deploy your apps directly without needing to set up a server. The host, Amazon RDS for example, manages all operating level stuff, meaning fewer overheads and quicker setup. PaaS is usually a time saver, though it has limitations, as the environment cannot be customised to the same extent as Iaas. There is no universal right or wrong approach – it depends on the app requirements. Only by understanding the app, the science that happens within it, and having been through the cloud migration process before, can you effectively identify the best approach.

A further challenge is migrating apps from one operating system to another. Cloud services often use the most cutting edge software, so an app moving from an old operating system to the cloud, will probably need some remediation to ensure it functions as before. This can be even trickier with small-medium applications for scientific purposes which are hosted on hardware outside the normal IT management processes of an organisation, meaning migration require skills to understand how unusual configurations can be best catered for in the new environment.

Companies moving to the cloud should also understand the tools available to do more with their apps and how to use them. Tools such as Amazon Machine Images for example allow you to take a snapshot of a server that you’ve set up in AWS and recreate it. This allows simple duplication and backup, but also allows you to experiment safely, and easily develop new test and proof of concept environments. Cloud starts to move your thinking away from viewing a server a permanent item you build once, to viewing it as a set of configuration you can reproduce and optimise over time, allowing you to choose more effective strategies as your understanding of the application grows.

Migrating these apps to the cloud benefits from hard-won experience. The complexity and specificity of the apps makes them a lot more challenging to migrate than common off the shelf apps. But experience of working with these and an understanding of the science that happens within them, as well as the role they play in the wider business context, allows approaches to be quickly identified which meet IT, business and user needs..

The result of getting it right is a greatly improved ability to run complex applications for processing and analysing complex scientific data, and so reach your business outcomes quicker. Large business going through an IT overhaul should take this opportunity to strategically migrate their scientific applications to the cloud.

Categorised:

James Grover

James Grover

James is a technical lead in the application continuity and improvement team, within which he has ...

Richard Brunt

Richard Brunt

Richard joined Tessella in 2015 after studying physics at Durham University. He now works as a ...

© Copyright 2017 Tessella
All rights reserved