AnyScale Applications
Lead Research Organisation:
University of Glasgow
Department Name: School of Computing Science
Abstract
The ecosystem of compute devices is highly connected, and likely to
become even more so as the internet-of-things concept is realized. There is
a single underlying global protocol for communication which enables all
connected devices to interact, i.e. internet protocol (IP). In this
project, we will create a corresponding single underlying global protocol
for computation. This will enable wireless sensors, smartphones,
laptops, servers and cloud data centres
to co-operate on what is conceptually a single task,
i.e. an AnyScale app.
A user might run an AnyScale app on her smartphone, then when the battery is running
low, or wireless connectivity becomes available, the app may shift its
computation to a cloud server automatically. This kind of runtime decision
making and taking is made possible by the AnyScale framework, which uses a
cost/benefit model and machine learning techniques to drive its
behaviour.
When the app is running on the phone, it cannot do very complex
calculations or use too much memory. However in a powerful server, the
computations can be much larger and complicated. The AnyScale app will
behave in an appropriate way based on where it is running.
In this project, we will create the tools, techniques and technology to
enable software developers to create and deploy AnyScale apps. Our first
case study will be to design a movement controller app, that allows a biped
robot with realistic humanoid limbs to 'walk' over various kinds of terrain. This
is a complex computational task - generally beyond the power of embedded
chips inside robotic limbs. Our AnyScale controller will offload
computation to computers on-board the robot, or wirelessly to nearby
servers or cloud-based systems. This is an ideal scenario for robotic
exploration, e.g. of nuclear disaster sites.
become even more so as the internet-of-things concept is realized. There is
a single underlying global protocol for communication which enables all
connected devices to interact, i.e. internet protocol (IP). In this
project, we will create a corresponding single underlying global protocol
for computation. This will enable wireless sensors, smartphones,
laptops, servers and cloud data centres
to co-operate on what is conceptually a single task,
i.e. an AnyScale app.
A user might run an AnyScale app on her smartphone, then when the battery is running
low, or wireless connectivity becomes available, the app may shift its
computation to a cloud server automatically. This kind of runtime decision
making and taking is made possible by the AnyScale framework, which uses a
cost/benefit model and machine learning techniques to drive its
behaviour.
When the app is running on the phone, it cannot do very complex
calculations or use too much memory. However in a powerful server, the
computations can be much larger and complicated. The AnyScale app will
behave in an appropriate way based on where it is running.
In this project, we will create the tools, techniques and technology to
enable software developers to create and deploy AnyScale apps. Our first
case study will be to design a movement controller app, that allows a biped
robot with realistic humanoid limbs to 'walk' over various kinds of terrain. This
is a complex computational task - generally beyond the power of embedded
chips inside robotic limbs. Our AnyScale controller will offload
computation to computers on-board the robot, or wirelessly to nearby
servers or cloud-based systems. This is an ideal scenario for robotic
exploration, e.g. of nuclear disaster sites.
Planned Impact
The ambitious nature and large scope of the AnyScale Apps project
means that it has the potential for major impact in areas of
academia, industry, UK economy and society.
Academic beneficiaries include the programming languages, software
engineering, computer systems, machine learning and robotics
communities. These are traditional research strengths in the UK,
with world-leading individuals and groups.
Many UK companies (and international companies with UK research
outposts) may benefit from this research project.
This range includes companies specializing in
app-development (e.g. Red Hat)
compilers (e.g. Codeplay)
middleware (e.g.Oracle Research Labs Cambridge)
operating systems (e.g. Microsoft Research Cambridge) and
processor architecture (e.g. ARM).
The UK ICT market is worth £140 billion, which is worth 12% of GDP.
AnyScale Apps could cause a profound impact on software development by pioneering a fundamentally
new way of developing and deploying software for heterogeneous scale
computing environments.
This would also allow developers to
harness advantages of cloud computing without tying down
to specific service provider, in effect, help in maintaining
computational independence.
The UK Cloud Computing annual market value is predicted to grow from £2.4 billion
to £6.1 billion by 2014.
Thus it is of immense value to UK economy to have skilled, trained software engineers
working on AnyScale apps and systems - those directly trained during
the project (RAs and associated PhDs).
AnyScale apps could help promote Green ICT by building notions of energy
efficiency into cost models. Just as electric appliances have EU energy
efficiency ratings A to G, software can also be made to have analogous ratings.
This in turn would allow society to adopt responsible (ethical) computation.
Longer term, the case study using robotics will be valuable
for autonomous exploration (e.g. nuclear disaster recovery)
and prosthetic limbs (e.g. rehabilitation after accident).
In these scenarios, bipedal robotic systems could be ideal - but
require large amounts of computation for adaptive movement.
Current systems are limited and would benefit from more intelligence due
to anyscale computation.
means that it has the potential for major impact in areas of
academia, industry, UK economy and society.
Academic beneficiaries include the programming languages, software
engineering, computer systems, machine learning and robotics
communities. These are traditional research strengths in the UK,
with world-leading individuals and groups.
Many UK companies (and international companies with UK research
outposts) may benefit from this research project.
This range includes companies specializing in
app-development (e.g. Red Hat)
compilers (e.g. Codeplay)
middleware (e.g.Oracle Research Labs Cambridge)
operating systems (e.g. Microsoft Research Cambridge) and
processor architecture (e.g. ARM).
The UK ICT market is worth £140 billion, which is worth 12% of GDP.
AnyScale Apps could cause a profound impact on software development by pioneering a fundamentally
new way of developing and deploying software for heterogeneous scale
computing environments.
This would also allow developers to
harness advantages of cloud computing without tying down
to specific service provider, in effect, help in maintaining
computational independence.
The UK Cloud Computing annual market value is predicted to grow from £2.4 billion
to £6.1 billion by 2014.
Thus it is of immense value to UK economy to have skilled, trained software engineers
working on AnyScale apps and systems - those directly trained during
the project (RAs and associated PhDs).
AnyScale apps could help promote Green ICT by building notions of energy
efficiency into cost models. Just as electric appliances have EU energy
efficiency ratings A to G, software can also be made to have analogous ratings.
This in turn would allow society to adopt responsible (ethical) computation.
Longer term, the case study using robotics will be valuable
for autonomous exploration (e.g. nuclear disaster recovery)
and prosthetic limbs (e.g. rehabilitation after accident).
In these scenarios, bipedal robotic systems could be ideal - but
require large amounts of computation for adaptive movement.
Current systems are limited and would benefit from more intelligence due
to anyscale computation.
Publications
Alnowaiser K
(2014)
A study of connected object locality in NUMA heaps
Cameron C
(2013)
A Virtual Machine for the Insense Language
Helfenstein J
(2022)
An approach for comparing agricultural development to societal visions.
in Agronomy for sustainable development
Cano J
(2018)
Automatic Parameter Tuning of Motion Planning Algorithms
Paredes M
(2016)
Breadth first search vectorization on the Intel Xeon Phi
Yiapanis P
(2015)
Compiler-Driven Software Speculation for Thread-Level Parallelism
in ACM Transactions on Programming Languages and Systems
Jalalinajafabadi F
(2015)
Computerised objective measurement of strain in voiced speech.
in Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual International Conference
Nikolaou N
(2016)
Cost-sensitive boosting algorithms: Do we really need them?
in Machine Learning
Sechidis K
(2017)
Dealing with under-reported variables: An information theoretic solution
in International Journal of Approximate Reasoning
Clarkson, J.
(2014)
Early Experiences In Embedding GPGPU Support In Java
Paredes M
(2020)
Exploiting Parallelism and Vectorisation in Breadth-First Search for the Intel Xeon Phi
in IEEE Transactions on Parallel and Distributed Systems
Mcpherson A
(2015)
Fence Placement for Legacy Data-Race-Free Programs via Synchronization Read Detection
in ACM Transactions on Architecture and Code Optimization
Kotselidis C
(2017)
Heterogeneous Managed Runtime Systems
Alnowaiser K
(2016)
Languages and Compilers for Parallel Computing
Brown G
(2015)
On unifiers, diversifiers, and the nature of pattern recognition
in Pattern Recognition Letters
Vanderbauwhede Wim
(2019)
Operating Systems Foundations with Linux on the Raspberry Pi: Textbook
| Description | Building, deploying and maintaining robust distributed systems is difficult! Use of virtual machines makes things somewhat more straightforward - the AnyScale Apps project involves allowing resource constrained devices to adapt their computation to meet available resource constraints. We have demonstrated this concept on a range of applications, including robotic motion planning, smart environmental sensors and programming language runtime systems. |
| Exploitation Route | Our resource-aware task scheduling algorithms might be useful in a variety of different contexts. For example, our smart sensor hardware with appropriate anyscale runtime support might be deployed in smart buildings - tests at the University of Glasgow campus are in operation at present. |
| Sectors | Digital/Communication/Information Technologies (including Software),Education,Environment,Leisure Activities, including Sports, Recreation and Tourism |
| Description | Raspberry Pi based sensor boxes have been deployed as part of the University of Glasgow's smart campus infrastructure, to perform indoor environmental sensing tasks. So far some tens of devices have been deployed, with a general room occupancy prediction service enabled. We intend to scale up to 250 devices, with more app-based student-focused services. Follow-on work on this framework has been funded internally at the University of Glasgow. Further sensor boxes are deployed as 'smart greenhouse' sensors and 'smart home' in various locations, in collaboration with the Urban Big Data Centre at Glasgow. Again, this is work in progress - along with a local SME we are working to make the smart sensor concept more user-friendly for end-users. |
| First Year Of Impact | 2017 |
| Sector | Communities and Social Services/Policy,Digital/Communication/Information Technologies (including Software),Education |
| Impact Types | Societal |
| Description | ACTiCLOUD - ACTivating resource efficiency and large databases in the CLOUD |
| Amount | € 4,733,532 (EUR) |
| Funding ID | 732366 |
| Organisation | European Commission H2020 |
| Sector | Public |
| Country | Belgium |
| Start | 01/2017 |
| End | 12/2019 |
| Description | Amazon Web Services University Research Scheme |
| Amount | $1,000 (USD) |
| Organisation | Amazon.com |
| Sector | Private |
| Country | United States |
| Start | 01/2016 |
| End | 12/2016 |
| Description | E2DATA - European Extreme Performing Big Data Stacks |
| Amount | € 4,676,250 (EUR) |
| Funding ID | 780245 |
| Organisation | European Commission H2020 |
| Sector | Public |
| Country | Belgium |
| Start | 01/2018 |
| End | 12/2020 |
| Description | Interface Standard Innovation Voucher |
| Amount | £5,000 (GBP) |
| Organisation | Government of Scotland |
| Department | Scottish Funding Council |
| Sector | Public |
| Country | United Kingdom |
| Start | 11/2017 |
| End | 03/2018 |
| Title | Glasgow Computing Science Environmental Sensors |
| Description | An integrated internet-of-things framework with a user-friendly web front-end to show current environmental status (temperature, light, sound) and room occupancy within public spaces in the university campus. |
| Type Of Technology | Webtool/Application |
| Year Produced | 2017 |
| Impact | Multiple paper submissions (in progress). Glasgow smart campus redevelopment potential. Interface UK innovation bid with networks4learning (ongoing) |
| URL | http://sensors.anyscale.org/ |
| Title | MR4J Optimiser |
| Description | A Java-based utility to support domain-specific parallel programming abstractions (the map-reduce project in beehive-lab). This utility adapts the underlying code at runtime to improve its efficiency. |
| Type Of Technology | Software |
| Year Produced | 2016 |
| Open Source License? | Yes |
| Impact | Publications arising from this work: Towards co-designed optimizations in parallel frameworks: A MapReduce case study. Colin Barrett, Christos Kotselidis, Mikel Luján. ACM International Conference on Computing Frontiers 2016 (preprint available at http://arxiv.org/abs/1603.09679). |
| URL | https://github.com/beehive-lab/map-reduce-optimiser |
| Title | MapReduce for Java (MR4J) |
| Description | MR4J is a MapReduce software component framework class building on standard Java parallel libraries to manage the scheduling of tasks in the different phases of execution on manycore processors. An optimiser is available to reduce the overhead associated with the intermediate data without extending the API. |
| Type Of Technology | Software |
| Year Produced | 2016 |
| Open Source License? | Yes |
| Impact | Publications arising from this work: Towards co-designed optimizations in parallel frameworks: A MapReduce case study. Colin Barrett, Christos Kotselidis, Mikel Luján. ACM International Conference on Computing Frontiers 2016 (preprint available at http://arxiv.org/abs/1603.09679). |
| URL | https://github.com/beehive-lab/map-reduce |
| Title | Tornado Virtual Machine |
| Description | The Tornado VM is a practical heterogeneous programming framework for automatically accelerating Java programs on heterogeneous (OpenCL-compatible) hardware. |
| Type Of Technology | Software |
| Year Produced | 2018 |
| Open Source License? | Yes |
| Impact | The system is in use for data analytics research projects at various UK universities. |
| URL | https://github.com/beehive-lab/TornadoVM |
| Title | Tornado Virtual Machine |
| Description | The Tornado VM is a practical heterogeneous programming framework for automatically accelerating Java programs on heterogeneous (OpenCL-compatible, NVidia GPUs and Intel OneAPI) hardware. |
| Type Of Technology | Software |
| Year Produced | 2018 |
| Open Source License? | Yes |
| Impact | TornadoVM has been used as the research platform of the following EU/UKRI projects: H2020 ELEGANT, EU Horizon/UKRI INCODE, AERO, TANGO, ENCRYPT. The total amount of funding from the research projects attributed to the University of Manchester is over £2M. In addition, TornadoVM is being funded by Intel as part of their strategy for OneAPI on how to program heterogeneous computing platforms using open standards. TornadoVM has been highlighted by Intel in several ways including the Intel 2021 Outstanding Researcher Award and DevMesh Spotlight. In addition, the University of Manchester has been invited to participate in the OpenAPI steering committee representing the only EU University in the consortium. Beyond research, TornadoVM has been presented in these industry-focused software development venues such as JVMLS, InfoQ, Devoxx, and others. TornadoVM is currently open source and is part of the incubation staged of the University of Manchester Innovation Factory for future commercialization. |
| URL | http://www.tornadovm.org |
| Title | incPy |
| Description | development/port of a compute-cache version of the Python 2.7 interpreter, based on earlier work from Stanford |
| Type Of Technology | Software |
| Year Produced | 2014 |
| Open Source License? | Yes |
| Impact | ongoing academic investigation / paper submissions |
| URL | https://bitbucket.org/davidrobertwhite/incpy |
| Description | ARM Research summit presentation 2016 |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | Three researchers from the AnyScale project (Lujan, Singer, Nagarajan) gave presentations about the research project, to an audience of ARM employees and other interested parties. This raised lots of questions and potential avenues for future research e.g. in robotics. |
| Year(s) Of Engagement Activity | 2016 |
| URL | https://developer.arm.com/research/summit/previous-summits/2016/speakers |
| Description | EdLambda user group |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | Regional |
| Primary Audience | Professional Practitioners |
| Results and Impact | 50 programming enthusiasts attended evening workshops about programming techniques, with presentations from AnyScale research team. |
| Year(s) Of Engagement Activity | 2015,2016 |
| URL | http://www.edlambda.co.uk/ |
| Description | Raspberry Pi Cloud |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Public/other audiences |
| Results and Impact | Many individuals and organisations across the world (USA, India, EU) are reproducing our innovative testbed for cloud computing Invitations to speak at Industrial and Academic events, national and international |
| Year(s) Of Engagement Activity | 2013,2014 |
| URL | http://raspberrypicloud.wordpress.com |
| Description | Research presentation (Glasgow) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Postgraduate students |
| Results and Impact | Clarkson presented the Tornado framework for GPU acceleration of Java code. |
| Year(s) Of Engagement Activity | 2016 |
| URL | http://www.dcs.gla.ac.uk/research/gpg/abstracts/2016/18-05-16.txt |
| Description | Researcher Lightning Talk (Glasgow) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | Local |
| Primary Audience | Other audiences |
| Results and Impact | Short presentation by project research assistant (Michala) about AnyScale research topic, providing information about an environmental sensing case study. The outcome was dissemination of information across various research groups, which led to new collaborations in further case studies. |
| Year(s) Of Engagement Activity | 2017 |
| Description | Talk at the MoreVMs workshop 2017 |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | A presentation about Virtual Machine engineering, to a mixed academic/industrial audience. |
| Year(s) Of Engagement Activity | 2017 |
| URL | https://2017.programming-conference.org/event/morevms-2017-papers-horizontal-profiling-for-virtual-m... |
| Description | User Group talks |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | Yes |
| Geographic Reach | Regional |
| Primary Audience | Professional Practitioners |
| Results and Impact | technical discussion, possible suggestions of industrial collaboration Ongoing interest in our open source codebase from industry |
| Year(s) Of Engagement Activity | 2013,2014 |
| URL | http://ukjugs.org |
| Description | Virtual Machine Users Group |
| Form Of Engagement Activity | Participation in an activity, workshop or similar |
| Part Of Official Scheme? | Yes |
| Geographic Reach | National |
| Primary Audience | Professional Practitioners |
| Results and Impact | invited talk on cloud computing to Scottish branch of VMUG established relationship with VMUG, invited to speak at future events, made contact with local companies regarding possible future collaboration (internships etc) |
| Year(s) Of Engagement Activity | 2014 |
| URL | http://www.vmug.org.uk |
| Description | Workshop talk (VM Meetup) |
| Form Of Engagement Activity | A talk or presentation |
| Part Of Official Scheme? | No |
| Geographic Reach | International |
| Primary Audience | Industry/Business |
| Results and Impact | Clarkson presented the Tornado framework for OpenCL acceleration of Java programs, to an audience of industry and academics. Several people expressed an interest in integrating with this open source framework. |
| Year(s) Of Engagement Activity | 2016 |
| URL | http://vmmeetup.github.io/2016/ |