About

The Project

Despite their proliferation as a dominant computing paradigm, cloud computing systems lack effective mechanisms to manage their vast amounts of resources efficiently, leading to severe resource waste and ultimately limiting their applicability to large classes of critical applications that pose non-moderate resource demands. This creates a significant need to lift existing technological barriers of actual fluidity and scalability of cloud resources towards promoting cloud computing as a critical cornerstone for digital economy.

ACTiCLOUD proposes a novel cloud computing architecture for drastically improved management of cloud resources, targeting both resource efficiency and scalability. By utilizing modest investments on hardware intelligence that enables true resource disaggregation between multiple servers, ACTiCLOUD progresses current state-of-the-art in hypervisors and cloud management systems, promoting holistic resource management both at the rack scale and across distributed cloud sites.

On top of this, ACTiCLOUD will evolve the ecosystem around in-memory databases, a core component for extremely demanding and critical classes of applications that up to now have faced severe difficulties in matching their resource requirements to state-of-the-art cloud offerings, with a final goal to provide cost-efficient and highly performant DataBase-as-a-Service (DBaaS) cloud platforms. To this end, ACTiCLOUD optimizes system software and managed runtime systems that are heavily utilized by databases and Big Data management frameworks and evolves two cutting-edge representatives of database frameworks, MonetDB and Neo4j.

Objectives

Embracing server platforms for resource disaggregation

ACTiCLOUD is implemented on top of two novel European server technologies (Numascale & Kaleao) that support resource disaggregation.

Implementation of a rack-scale hypervisor

ACTiCLOUD develops a rack-scale hypervisor to undertake the critical role of resource virtualization for the unified pool of resources that are augmented from multiple physical servers and allocated to different applications in a multi-tenant environment.

Enabling holistic resource management in distributed clouds

To better utilize all the available resources, distribute load according to specified policies, and enable dynamic decision making and actions, ACTiCLOUD implements a hierarchy of scheduling models at the rack, site and cross-site levels.

Optimization of system software & language managed runtimes

Towards activating large databases in cloud ecosystems, ACTiCLOUD optimizes substrate software libraries and runtime systems that also significantly affect the management of resources of the upper database layers.

Cloudification of in-memory, column-store databases

With its in-memory optimized technology, MonetDB focuses on data-intensive data warehouse analytics, requiring scale-up to 100s GiBs of memory and tens to hundreds of processing cores. Through ACTiCLOUD, MonetDB advances from a static DBMS to a dynamic elastic cloud DBMS, by strengthening its scalability in both directions, i.e., vertical and horizontal, and by quickly reacting to workload changes dynamically increasing or decreasing resource allocation.

Cloudification of large-scale graph databases

Neo4j is the world's first and leading graph database. Through ACTiCLOUD, Neo4j gets equipped to process graphs that far outstrip main memory constraints and bring many cores into the analysis.