Hardware-Assisted Decoupled Access Execution on the Digital Market: The EDRA Framework

Project Information

EDRA

Grant agreement ID: 851631

DOI

10.3030/851631

Project closed

EC signature date 1 April 2019

Start date 1 May 2019

End date 31 October 2020

Funded under

• EXCELLENT SCIENCE - Future and Emerging Technologies (FET)

Total cost

€ 100 000,00

EU contribution

€ 100 000,00

Coordinated by

EREVNITIKO PANEPISTIMIAKO INSTITOUTO TILEPIKONONIAKON SYSTIMATON
Greece

Periodic Reporting for period 1 - EDRA (Hardware-Assisted Decoupled Access Execution on the Digital Market: The EDRA Framework)

Reporting period: 2019-05-01 to 2020-10-31

Summary of the context and overall objectives of the project

EDRA focuses on the commercialization of the DAER technology developed during the FET H2020 project EXTRA (Exploiting eXascale Technology with Reconfigurable Architectures - GA 671653). The project targets the cloud acceleration market, a sector valued at 3.5 B$ in 2019, and estimated to increase approximately 10 times by 2025, reaching up to 35 B$. EDRA will use an in-house framework to offer a DAE-as-a-Service (DaaS) to customers for fast deployment of application-specific IPs on FPGA-accelerated cloud machines.
EDRA will enable access to HPC platforms with low investment risks, the key for economic growth and business expansion. The project will allow developers to improve application execution with minimal cost to hardware-assisted cloud infrastructure. SMEs and companies will have the option to enable the creation and adoption of new services, addressing a wide-range of potential customers, and ultimately increase their growth and expand their market share.
As shown in Figure "edra_steps_towards_framework.png", the project objectives are:
1. Build the AWS compatible software library stack, which hides all complex software tasks for interfacing the hardware (e.g. code transformation, DAE-compatible executables, transparent CAD tool invocation for generating / updating the hardware). Overall, the final EDRA software stack will expose a lightweight developer API for easy and straightforward application deployment to the AWS marketplace.
2. Deploy the EDRA service on the AWS marketplace.
3. Maximize market growth and profitability, by ensuring that the EDRA framework becomes available to as many customers as possible, supported by an efficient charging strategy for optimal profitability.
Eventually EDRA focused also on the development of a Minimum Viable Product (MVP) that accelerates phylogenetics analysis on Amazon’s FPGA-supported machines. The MVP can be easily launched within a few minutes from biologists, offering up to 2.5x better performance, compared to software-optimized solutions.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

EDRA has successfully met its goal on making the DAER technology ready for entering the market by reaching the following achievements:
• EDRA framework: The project successfully updated the EXTRA IP, and developed a full-fledged framework that can (semi) automatically create DAE-compatible applications, capable to facilitate available hardware resources for faster workload processing.
• MVP in the domain of phylogenetics: The EDRA team has strong scientific background in the domain of accelerating workload processing related to phylogenetics using FPGAs. Based on this fact, the team decided to develop and deploy a first MVP using the EDRA framework that accelerates phylogenetics analysis on Amazon’s FPGA-supported machines.
• MVP market analysis: Towards validating the decision of deploying the MVP to the market, the team conducted a thorough analysis on market size and opportunities with respect to the bioinformatics domain. Results suggest that an estimated market size directly fitting to the EDRA MVP is valued at approximately 7.9 M€, 10.8 M€ and 14.6 M€ for 2021, 2022 and 2023 respectively.
• Business model formulation: EDRA developed a complete and sustainable business model. Its value proposition is based on solutions to customers who would like to execute applications faster compared to their current setup, as well as reduce IT costs related to deployment and maintenance. Due to the team expertise on hardware design and background on phylogeny, EDRA plans to also provide dedicated application acceleration services on phylogenetics. Key resources required to support the EDRA value proposition, are budgets related to staff support, cloud / digital resources, facilities and IPR management. Customer segments comprise users from the HPC domain, as well as academic institutes in the domain of bioinformatics. As shown in Figure “edra_ami_aws.png”, the revenue model is based on fixed charges for deploying customer applications to Amazon’s marketplace using the EDRA framework. The EDRA’s MVP revenues will be based on Amazon’s pay-as-you-go charging policy.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

EDRA successfully delivered an end-to-end framework for mapping applications onto Amazon’s FPGA-supported cloud platforms. Moreover, the project deployed a pioneering MVP that enables faster processing of workloads related to phylogenetics. This research domain is crucial for understanding organism evolution, and especially of viruses like the SARS-CoV-2 and its mutations. As such, the project socio-economic impacts with respect to its developed technology and achievements are the following:
• Economic and business impact: The project has delivered the EDRA framework, a genuine technology for rapid and (semi) automatic hardware-accelerated deployment of applications to cloud resources. The framework allows essentially the quick launch of cloud-supported services for SMEs and corporations, thus enabling faster and better services for end users, a key aspect for economic growth.
• Increased value creation from FET projects by picking up innovation opportunities: Based on the framework developed by EDRA, the team formulated a go-to-market strategy for offering hardware-acceleration services to the cloud for demanding applications. Moreover, EDRA released an MVP that enhances research on phylogenetics, an important area that can assist on the fight against the COVID-19 outbreak and other potential pandemics.
• Improved societal and market acceptance: The COVID-19 outbreak demonstrated that scientists need access to powerful computational resources with fast turnaround times of results on drug analysis and model simulation. EDRA’s MVP addresses today’s the important need for more computational power towards faster analysis of virus evolution.
• Contributing to the competitiveness of European industry/economy: EDRA picked up on the fact that major cloud providers, such as Amazon and Alibaba, started offering services that support FPGAs; the DAE framework allows the offer of generic low-risk hardware-acceleration services on the cloud for customers that wish to remove their application back end from their premises.
• Stimulating, supporting and rewarding an open and proactive mind-set: EDRA’s ability for quick deployment of hardware-accelerated applications, allow SMEs and corporations to investigate and propose new services to end-users with minimum investment risk, strengthening even more the European industry sector.
• Scientific impact: EDRA’s MVP is a genuine solution that allows biologists and researchers in phylogenetics to increase their productivity, while at the same time reducing IT costs. The majority of the software tools used for experiments are compute-bound, hence the additional processing power that EDRA’s MVP provides can further assist on understanding the origins of lethal viruses. This is a valuable asset for constraining the spread of potential new pandemics, should anytime happen.
Finally, EDRA has already initiated the process of launching a spin-off towards entering the market based on the formulated business plan. The spin-off plans to deploy the MVP to Amazon’s marketplace, where biologists and researchers on phylogenetics will be able to download it on a pay-as-you-go charging policy, and instantly conduct their experiments up to 2.5x faster compared to the currently available software-optimized configurations.