Have you ever wondered what happens if you put Europe’s finest cloud computing researchers in one place for three days? So have we. This is why we are joining the Cloud Forward 2015 Conference in Pisa. We are excited to discuss the future of computing with European thought leaders in all things cloud.
We are also there to represent MIKELANGELO in discussions. From those discussions we expect to gain fresh insights how to position MIKELANGELO in the light of bleeding edge research. We will use these insights to engineer a future-proof cloud, agile services, and generate more value for you.
The conference’s tag line is “From distributed to complete computing”. But what does that mean? How does MIKELANGELO fit in? Let’s find out together.
About the Cloud Forward 2015 Conference
The Cloud Forward 2015 Conference takes place October 6th-8th in Pisa. HOLACloud, a Horizon 2020 project like us, have organised Cloud Forward 2015. HOLACloud analyses trends in cloud computing and big data, it publishes a journal, and fosters the collaboration between cloud computing researchers in Europe. Cloud Forward 2015 is organised in this spirit of European collaboration.
You can see the most important keywords from the conference programme in the tag cloud below. The striking keywords cover cloud computing, its challenges, and new approaches to distributed computing.
Representatives from academia, industry, and the European Commission chair the sessions and give expert talks. The conference’s theme, “From distributed to complete computing”, is reflected in the conference programme by parallel sessions on distributed computing and clouds on one hand, and complete computing on the other.
Now you might ask: “What exactly is complete computing”? We have to admit; we don’t know either. And apparently neither does Google. We conclude that the topic must be really bleeding-edge. The more important for us to dig deeper.
According to the call for papers the track on complete computing covers the topics
- “Experiences of utilising products from cloud computing research efforts”,
- “Problems encountered in utilising cloud computing”,
- “Requirements of ICT systems beyond cloud computing”,
- “Innovative solutions beyond cloud computing”.
The resulting conference sessions on complete computing contain titles such as:
- “Beyond Mere Application Structure: Thoughts on the Future of Cloud Orchestration”,
- “Automated deployment of a micro-service-based monitoring infrastructure”,
- “The EGI Federated Cloud e-Infrastructure”.
These topics show that the conference aims to go beyond cloud computing. The topics also show it is time to leverage our experiences in cloud computing and to move to a more flexible, higher level of abstraction.
MIKELANGELO at Cloud Forward 2015
Our technical coordinator, from XLAB, will represent MIKELANGELO at Cloud Forward 2015. He will join the session on “New approaches for infrastructure services”, chaired by Dana Petcu, Professor at the Computer Science Department of West University of Timisoara. In this session, Gregor will present and discuss our project goals and our approach to a more agile and robust cloud. These discussions will drive our project’s agenda to offer you a better cloud experience.
Now you wonder what MIKELANGELO is all about? I’m glad you asked!
MIKELANGELO is about resolving the most pressing issues of virtual infrastructures. We work hard to improve I/O performance and security of virtual machines, to provide comprehensive and holistic monitoring, and to integrate hassle-free application packaging and deployment. Our main areas of work are shown in the architecture diagram below.
In the diagram you can see that we work on many technologies. One of our strengths is a tight integration between the base technologies that make up the cloud and HPC. Our preferred technologies are printed in coloured font. Having preferred components does not mean that we don’t play along nicely with others, though. Each new and modified component works with legacy systems as well. But our preferred mix will give you the most benefits over legacy systems.
Cloud computing is maturing; and so are its base technologies. High performance computing is even much farther along the maturity curve. Thus, we approach the topic by extending proven cloud and HPC technologies.
We want to smoothen the rough edges of great technologies and integrate new breakthrough concepts. This approach will make the fruits of our work available in no time instead of having product release cycles of a couple of years.
Our architecture diagram shows that our work touches most software layers of modern computing paradigms. Let’s quickly go through them bottom-up.
In the virtualisation layer, we extend KVM by dynamic management of I/O cores in a component we call IOcm. To gain even more I/O performance, we add vRDMA capabilities to the hypervisor. vRDMA makes inter-VM communication much faster by reducing communication overhead. sKVM, our modified version of KVM, integrates IOcm for dynamic I/O core management and vRDMA.
Our preferred guest operating system in virtual machines is OSv. OSv is built to be a guest operating system in the cloud. One of its major benefits is a tiny footprint. Thus, OSv allows us to use infrastructure resources much more efficiently. More efficiency leads to improved performance and reduced costs.
Besides OSv, of course, we also support traditional cloud guests such as Ubuntu and CentOS. This means that you can gain benefits from our computing stack with your existing virtual machine images.
We leverage sKVM and OSv by integrating with Torque for high performance computing and OpenStack for cloud computing. For both paradigms we implement novel security mechanisms and holistic monitoring.
From a bird’s eye view, our work covers pretty much the whole software side of the modern computing stack. At the same time we make incremental adoption a breeze. Most components can be integrated in legacy systems individually.
To show you that we are serious and that our promise actually holds up, we demo our stack in four use cases. These use cases stress the benefits of our computing stack in realistic scenarios. They span big data, high performance computing, OpenFOAM and cloud bursting.
You can learn more about our approach and our use cases by reading our most recent reports and use case descriptions.
We are excited to exchange ideas about future virtual infrastructures at Cloud Forward 2015. We are also excited to promote future virtual infrastructures based on MIKELANGELO in new areas such as high performance computing.
All in all, we see MIKELANGELO as potential driver of pervasive, integrated, and highly performing virtual infrastructures. One could probably call it “complete computing”.