Modelling and Orchestrating heterogeneous Resources and Polymorphic applications for Holistic Execution and adaptation of Models In the Cloud

Project Information

MORPHEMIC

Grant agreement ID: 871643

DOI

10.3030/871643

Project closed

EC signature date 24 November 2019

Start date 1 January 2020

End date 30 June 2023

Funded under

• INDUSTRIAL LEADERSHIP - Leadership in enabling and industrial technologies - Information and Communication Technologies (ICT)

Total cost

€ 4 988 693,75

EU contribution

€ 4 988 693,75

Coordinated by

UNIVERSITETET I OSLO
Norway

Periodic Reporting for period 2 - MORPHEMIC (Modelling and Orchestrating heterogeneous Resources and Polymorphic applications for Holistic Execution and adaptation of Models In the Cloud)

Reporting period: 2021-07-01 to 2023-06-30

Summary of the context and overall objectives of the project

The main objective of MORPHEMIC is to simplify Cloud application modelling and continuously optimize and morph the deployment model to take advantage of beneficial Cloud capabilities.
Cloud computing offers the possibility of utility computing on demand and may help Small and Medium sized Enterprises (SMEs) to run their software applications beyond the capacity of their own infrastructures. However, renting computing resources will be more expensive than operating own infrastructure in the long run, and Cloud computing is therefore ideal for short lived applications or applications whose resource requirements vary over time or where extra effort and computational bursts are required to meet computational deadlines. It is currently demanding to run applications across multiple Cloud providers, and costly to constantly monitor and manually scale the application of long duration to remove bottlenecks and ensure the continuous satisfaction of the application users.

MELODIC responded to this challenge by offering multi-Cloud optimization platform for automatic deployment and application management of the application to different cloud providers in a fully Cloud agnostic approach. The DevOps engineer describes the application in the Cloud Application Modelling and Execution Language (CAMEL), including the operational goals and the operational constraints, and then MELODIC takes care of the deployment and application reconfiguration as the application execution context changes. For instance, in response to an upsurge in the number of users. MELODIC reacts with corrective application reconfigurations to changes in the context, and it can only take actions to satisfy the goals of the CAMEL model but not modify the model itself to make the application benefit from new hardware or Cloud capabilities available.

MORPHEMIC is a pre-processor for MELODIC helping the DevOps engineer to create a better CAMEL model from the start, and proactively modify the CAMEL model in parallel with MELODIC managing the Cloud application at run time. MORPHEMIC is thus an outer feedback loop continuously looking for a more optimized CAMEL model. This will be done with minimal changes to the MELODIC platform, but using the information gathered by MELODIC and machine learning to understand what the better CAMEL model for the managed Cloud application will be.

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

CAMEL has been extended to support the modelling of polymorphic applications. The extensions covered various domains of CAMEL like deployment and metric. The Web Crawler and the Knowledge Base were implemented and released, that provides the Profiler with functionalities of metadata search, storage, and filter. The components can interact with the other services of the MORPHEMIC platform. The CAMEL language and supporting tools have been further improved based on the feedback from the end users during the last project period, and work on the code classification is in progress.
MORPHEMIC has devised a methodology whereby the DevOps engineers can formulate marginal utilities as unary functions individually over each measured metric of the application’s execution context. The idea is that the DevOps engineer can more intuitively relate the evolution of a single measurement value to the utility of the deployment.
The end-to-end architecture of the MORPHEMIC Pre-processor and proactive adaptation has been designed. The overall architecture, with justification of architecture decision has been provided, as well as plan for the detailed design preparation for each component. The security architecture of MORPHEMIC has been produced and documented.
Cloud offering collection and matching techniques have been developed to support candidate node filtering. The most prominent monitoring tools across 10 different criteria were analysed comprehensively.
A competitor analysis has been performed, and although this is continuous work, the analysis reported so far shows that MORPHEMIC has a unique value proposition to possible clients. An extensive publicity and awareness campaign has been designed and is currently ongoing, and several industrial talks and demonstrations have been undertaken. Currently MORPHEMIC has been presented on over 50 industrial and open-source conferences on almost all parts of the globe. The initial cooperation with the potential users of the MORPHEMIC platform outside of the consortium has started. Two companies are heavily interested in using MORPHEMIC. There will also be applications for national funding from several partners, and ongoing project proposal initiatives using. The research activities performed in MORPHEMIC has been presented to several on-going H2020 projects (RAINBOW, PIACERE, AI SPRINT) with the potential for joint research on these topics. MOPRHEMIC team members are very active in H-CLOUD initiative sharing their work and results among the projects included in that initiative.

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)

By the end of MORPHEMIC we expect to have a novel theory for systems controlled based on data forecasted from measurements of the controlled systems itself. We will also have a complete theory for utility modelling for Cloud deployment and application management aiming to reduce significantly the required mathematical knowledge of the system owner and building the utility assessment on high level goals and assessments of the available application monitoring measurements. We will also push the boundaries for application monitoring to a multitude of devices and build dynamic support for hardware accelerators to allow more computationally cost-effective applications. A new version of the CAMEL modelling language and supporting modelling tools will make it easier for the application owners to exploit the benefits of autonomic cross Cloud application deployment and management. This will avoid that the application owners are being locked into commercial relationships with one and only one Cloud provider and its offerings.

Small and Medium sized Enterprises (SMEs) will benefit from the MORPEHMIC platform taking care of application management tasks that would otherwise require specially trained personnel on duty around the clock. MORPHEMIC, being open-source software with professional commercial support, will allow them to use MORPEHMIC platform freely and hopefully contribute to maintain and improve MORPHEMIC in the future. Furthermore, the support for special made hardware will also allow SME applications to benefit from the power of hardware driven artificial intelligence, dedicated signal processing, and integration with the Internet of things. This will enable companies to do more, better, and faster than today. This again can help European companies to use Cloud computing for challenging tasks like complex simulations, experiments, and data analytics. These are important areas where small companies today may have ideas for new products and services but must follow a more costly path to the market without the support of powerful and scalable computing solutions offered by the Cloud. MORPHEMIC will lower the barriers to use eScience and dScience for everyone on a freely chosen to mix of Cloud providers and computing platforms to suit the time variate resource needs of long running and demanding applications.

MORPHEMIC research and technical works and information flow