Taking the middleware out of high performance computing
Harnessing the full power of cloud computing can be difficult for developers as cost, complexity and time constraints often get in the way. Adding to this dilemma is the fact that traditional high performance computing relies on excessive amounts of middleware, orchestration and over engineering.
To make it easier for developers to create distributed applications which can solve large, complex and computationally intensive challenges, Hadean has created a new distributed computing platform that allows them to write scalable cloud-native applications.
To learn more about Hadean and its new platform, TechRadar Pro spoke with the company’s VP of operations, Mimi Keshani.
Can you tell us about how you came to found Hadean and the problems you set about trying to solve?
Hadean was founded to overcome the challenges posed by distributed computing and unleash the potential of the cloud. Taking a first-principles approach we developed the Hadean platform, which enables programmers to quickly and easily write networked, dynamically-scaling cloud-native applications.
How does Hadean work to overcome computational bottlenecks?
Hadean removes computational bottlenecks by eliminating excessive middleware, orchestration and overengineering, enabling developers to build, ship and scale their applications quickly and cost-effectively.
Hadean sits close to the metal and implements a unique process model that transforms the reliability and scalability of distributed computing, ensuring any application built on it is distributed and dynamically scalable by default.
The Francis Crick Institute recently announced it is working with your company to simulate how Covid-19 spreads in the body and through the population. What has Hadean’s role in this collaboration looked like?
Our simulation application, Aether Engine, will be used by the Francis Crick Institute to carry out COVID-19 simulations of models that combine analysis of person-to-person interaction with insight into how the virus transmits within an individual, providing a multiscale picture of the pathogen’s spread. The project will involve computationally complex data sets, which will be handled by Aether Engine’s distributed octree data structure. It dynamically partitions the simulation to provide additional computing power as required, reducing the expense and engineering complexity typically associated with running hyper-scale simulations.
How will governments and health organizations benefit from the large-scale models developed by your company’s solution? Were any specific findings novel or unexpected?
Ultimately, we hope to accurately predict how susceptible an individual is to infection and the likelihood that they can pass it on. The project will provide near real-time analysis, and enable governments and health organisations to make more informed decisions when planning protective measures, which will ultimately ease the burden on the healthcare infrastructure and save lives.
What is your take on the surge in cloud adoption as a result of the pandemic? Do you perceive any potential issues?
Technology is in a constant state of rapid evolution, and the recent pandemic has highlighted the need for greater computing power, and the ability to access it quickly and cheaply. In pandemics we also see a proliferation of data becoming available, and therefore greater demand for cloud technologies to store, structure, and analyse these outputs.
Unleashing the potential of both big data and big compute means that we are now able to rigorously explore avenues such as personalised and precision medicine. In order to make the most of the computation power available we need to build cloud-native applications – hence why we created Hadean.
Your company worked on a previous project involving protein-protein interactions last year. Can you tell us a bit more about that project and how it led to your work with the Francis Crick Institute?
In 2019 we partnered with the Francis Crick Institute to explore Hadean’s applicability to the life sciences via an Innovate UK grant. We used Aether Engine, our spatial simulation engine, to investigate how we could accelerate computational tools for docking two protein structures together, including a number of interactions involved in certain cancers.
We were able to significantly reduce the computation time required to sample millions of possible protein structures – as sampling on Aether Engine can be easily parallelised with no extra effort for the developer, and using more diverse inputs meant fewer docking runs were needed. We published this work in the journal Proteins and renewed the contract to build on this work and model COVID-19 transmission.
Can you tell us a little more about the Aether Engine and how you believe it could be used in future research projects?
Aether Engine is a spatial simulation application. It scales across different processors and physical machines, utilising more computing power as the simulations grow in complexity and size. It can run complex simulations, quickly and at massive scale; drastically improving the speed, scalability and reliability of cloud and distributed computing systems.
Can you tell us about HadeanOS, which you describe as the world’s first cloud-native operating system?
The Hadean Platform (formally OS) is a cloud-native operating system, which exists to implement the distributed process model. Applications built on the platform are distributed by default, without the need for containers, such as our cardinal application Aether Engine. The Hadean platform dynamically scales as more or less computing power is required by efficiently spliting a computational task and allocating it out to CPUs in a given cloud system, rather than that of a single server or cluster.
Are there any future projects or products your team is currently working on?
We’ve got a number of exciting projects in the pipeline. Most recently we’ve been helping epidemiological researchers at Imperial College via the RAMP initiative on their work to create massive spatial social networks of the UK to inform models being used to develop the NHSX contact tracing app.