Platform helps CERN build private cloud infrastructure

The European Atomic Energy Research Institute (CERN), founded in 1954, is the world's largest particle physics research center. CERN, with its powerful computing power, has been providing a solid guarantee for more than 17,000 researchers worldwide to work together to crack down on the mysteries of matter and the universe in 270 research centers across 48 countries.

To further accelerate research, CERN has started building the LHC, the world's largest Hadron Collider. The LHC is located in the vicinity of Switzerland and France, near the 100 meters underground, about 27 kilometers in the ring tunnel, its physical experiments on the basic theory of particle physics test and explore new breakthroughs, play a crucial role. In particular, the exploration of the origin of the universe and the search for new particle Higgs may result in a major breakthrough in particle physics, which leads to a new stage in the understanding of the micro-world, beyond the new physical laws of the standard model.

Most of the applications in the LHC project are highly specialized and require very good continuity, which poses new challenges to the management of the cluster of computers that support the project. The new cluster of computers that support the LHC project consists of 200 servers, but the overall utilization of the new clusters is only about 10% per cent, says Dr Helge Meinard, head of the research center, which handles the LHC data: "In launching projects like the LHC, We must ensure that all computing power is fully utilized. "To improve server utilization, CERN virtualized servers." The new virtualization cluster supports a variety of virtual machine management programs such as Hyperv and Xen, which gives the lab team more flexibility to choose the best platform for their applications, but at the same time, the use of multiple virtualization technologies adds management complexity to the entire cluster environment. CERN decided to use the Platform ISF solution to manage the LHC cluster workload as a tool for managing both virtual and physical environments to achieve the goal of building cost-effective, managed virtual machine management programs and operating systems, and shared computing infrastructure on any server.

"We finally chose platform as our partner to help us expand and optimize our unique high-performance computing environment, and to build an ideal platform for our entire scientific team to discover and explore the mysteries of the universe." "CERN has been using Platform LSF to manage workloads since 1997, Platform LSF also applies to the LHC," said Dr. Tony Cass, head of the CERN structural infrastructure (fabric infrastructure) group. Especially when you need to add cloud functionality, the platform ISF with platform LSF from the same vendor looks perfect. ”

As a cloud infrastructure management platform for centralized management of heterogeneous virtual and physical resources, Platform ISF provides a resource management and scheduling platform for rapid deployment, dynamic resource scheduling, and resilient computing for various applications running on it. It provides IAAS capabilities to help customers implement self-service on-demand computing, networking, storage resource applications, and usage. More importantly, it also makes middleware and distributed applications more convenient, migrate quickly to a cloud platform and have the ability to provide PAAs (platform-as-service) and SaaS (software-as-services) to ensure that the providers of middleware and distributed applications are more focused on their application logic without having to consider the underlying computing, storage, network resources scheduling and management.

Platform ISF provides a full range of private cloud computing infrastructures for a large number of scientists at CERN, providing efficient workload application management for multiple virtual and physical platforms. Platform ISF combines different platforms into a single dynamic shared infrastructure that dramatically improves platform utilization with less resources. In addition, scientists can choose their own application environment, dynamic control projects, to ensure that they can fully develop the effectiveness of the private cloud, all of which are in the unprecedented low cost state of implementation.

In addition, Platform ISF gives CERN greater control over the cluster environment, and automates management of many tasks by automating the hypervisor, simplifying management and improving management efficiency. Platform ISF can manage resources and application environments and provide contractual interfaces that allow users to reserve and use resources as needed.

Platform ISF's independence is of interest to CERN, Cass said: "CERN's scientific environment attaches great importance to the tools to adapt to the new development situation, rather than let us be bound by the old situation, Platform ISF done." ”

CERN deploys platform ISF to a cluster that supports the LHC, and deploys platform ISF re-use cluster to a HPC cluster of 3800 multi-core servers, which is responsible for batch processing, allowing CERN to use the private cloud computing environment , sharing resources between different clusters when necessary. Cass said: "Platform ISF re-use cluster can be combined with existing Platform LSF grid workload management solutions to provide our users and research centers with the scalability and flexibility needed to manage the cluster with shared data center resources, It also meets our requirements for open standards. ”

Platform ISF also helped CERN reduce the number of dedicated LHC cluster servers and ease the management burden. "In the past, in a static resource environment, the use of Platform LSF can reduce the workload of managing day-to-day applications, and now Platform ISF enables us to more effectively manage the mix of virtual machines and physical machines by transforming time-consuming human management into automated provisioning and management." Dr. Cass further explained: "By increasing server utilization, if you can reduce 150 servers from the existing LHC environment (originally 200 servers), then we can significantly save power and cooling costs, and without increasing the hardware budget, Redeploy the server to a batch cluster; If you can use the automatic configuration server capabilities provided by platform ISF to meet the workload requirements of the lab team, it will reduce our management burden. Cass predicts that with the use of platform ISF, CERN is expected to drastically reduce the amount of work required to reconfigure the batch cluster to meet changing workloads.