It’s just fascinating to watch the pace at which technologies are evolving. The latest Cisco Global Cloud Index shows just how fast everything in the cloud is growing.
- Annual global cloud IP traffic will reach 5.3 zettabytes by the end of 2017. By 2017, global cloud IP traffic will reach 443 exabytes per month (up from 98 exabytes per month in 2012).
- Global cloud IP traffic will increase nearly 4.5-fold over the next 5 years. Overall, cloud IP traffic will grow at a CAGR of 35 percent from 2012 to 2017.
- Global cloud IP traffic will account for more than two-thirds of total data center traffic by 2017.
In the HPC community – there is a need to process more information faster and a lot more accurately. In this whitepaper from Bright Computing, we learn how delivering a cost-effective HPC solution to Cisco Unified Computing System (UCS) customers has become a lot easier. Technologies like Bright Cluster Manager are a powerful software solution for managing HPC clusters cost-effectively. From that same perspective – Cisco UCS is a highly successful integrated server solution that offers a wide range of capabilities from a single vendor.
The true power shows when you use these technologies together. As the paper outlines, these products provide an HPC cluster solution that offers the following cost-saving features:
A single, high-performance Ethernet network fabric that can be used for all management and communications within the cluster.
- Top500 HPL benchmark performance that provides 89.69% efficiency across 512 Intel Xeon cores using Cisco Ethernet. Compared to Dassault Systems Abaqus benchmarks for QDR InfiniBand, Cisco UCS performed as well as or better than InfiniBand.
- The Cisco User-Space NIC (usNIC) provides 2.2μs latency when OpenMPI and a Cisco switch are used. Using various usNIC features, OpenMPI can create dedicated virtual hardware queues for highly efficient message-passing performance.
- Cisco UCS is a vertically integrated platform optimized by a single vendor that provides a unified fabric supporting three network types—interconnection, storage, and management.
- With the introduction of Bright Cluster Manager 7, any UCS C-Series server model can be easily and seamlessly managed as an HPC resource.
- Bright Cluster Manager provides deep and comprehensive control and manageability for deploying and managing UCS C-Series servers.
- Bright Cluster Manager provides full configuration and management of the Cisco usNIC Ethernet and Fibre Channel virtual interface.
- There is no need for specialized expertise to manage high-performance interconnects. Managers can utilize their in-house Ethernet expertise to build and operate an HPC cluster.
- A single, standard converged Ethernet fabric can easily be deployed to fit almost any storage scenario, including NAS, SAN, and FC.
Furthermore, with the introduction of Bright Cluster Manager 7, any Cisco UCS C-Series server model can be easily and seamlessly managed via API calls. It can configure, monitor, and manage C-Series rack servers directly in standalone mode or via the UCS Manager. Cisco Integrated Management Controllers (CIMCs) can be configured, controlled, and provisioned directly from Bright Cluster Manager without the 6200 Series embedded software UCS Manager.
The combination of Bright Cluster Manager and Cisco UCS servers provides an entirely new level of integration, management, and performance previously unavailable in the HPC market. The highly desirable all-Ethernet HPC cluster has become a viable solution for technical computing.
Download this whitepaper today to learn how a system built on Cisco UCS using Bright Cluster Manager is a complete HPC Ethernet solution that frees the IT manager, administrator, and users to focus on HPC results instead of managing a complex collection of hardware and software.
Get this and other HPC white paper from the insideHPC White Paper Library.
