September 13, 2012
IaaS provider ProfitBricks is hoping to raise some eyebrows with some recent benchmark results. The Berlin-based company commissioned Cloud Spectator, an IaaS benchmarking service, to compare its performance against Amazon and Rackspace. Given the outcomes, it appears that ProfitBricks is positioning itself as a viable service provider for HPC workloads.
According to their performance report, the company believes that infrastructure performance is primarily determined by processor, storage and interconnect capabilities. ProfitBricks US and German datacenters follow this mantra by running AMD bulldozer servers with an InfiniBand backbone. The interconnect design is rather impressive with servers running dual QDR cards, supporting up to 80Gbit/s connectivity.
"We deconstructed cloud infrastructure down to its most basic elements and discovered there was a far better way to deliver the service… Our platform allows ProfitBricks to offer unprecedented services at prices others cannot touch because of legacy design and built-in costs," said Achim Weiss, ProfitBricks chief executive officer, in an official statement.
To determine each IaaS provider's capabilities, Cloud Spectator ran a number of open-source benchmark tests including UnixBench, DBENCH, Iperf and Apache Kernel Compilation. The study took place over a two-day period in July and gauged server, CPU, storage and local network performance.
This test determined the overall performance of a system. Scores can be affected by both hardware and software, including the OS, compiler and libraries. When running UnixBench, ProfitBricks posted and average score of 1,567 vs. 1,031 and 948 for AWS and Rackspace respectively.
The Apache compilation test records how much time is required to build an Apache HTTP server. ProfitBricks was the only contender to complete the task in less than 1 minute (56 seconds). Amazon created the server in 61 seconds and Rackspace finished in 68 seconds.
DBENCH is an application that performs a file system stress test. The program returns the amount of concurrent programs or clients a file system can handle before experiencing heavy latency. Again, ProfitBricks came in ahead at 643. Both Rackspace and Amazon were far behind, scoring 164 and 77.
Iperf or Internal network performance, tests a network's throughput using TCP and UDP streams. Cloud Spectator ran these tests between two servers at the same datacenter. As expected, ProfitBricks took an exponential lead in this test, which is due to the InfiniBand infrastructure. It scored 5,754 vs. 397 and 289 for Rackspace and Amazon.
The company has also just announced real-time vertical scaling of CPU cores and RAM - with no reboot required. Until now, customers have had to preselect server sizes and add new servers as needed, but the new approach enables the selection of server instances from 1 CPU core to 48 cores, and from 1GB of RAM to 196 GB of RAM on the fly. All CPU cores are dedicated.
Given ProfitBricks' hardware and dedication to low-latency, high-throughput interconnects, it's somewhat surprising that GPU computing or bare metal services are not being offered. If those features are made available in the future, and are offered at the right price point, the IaaS provider will have a comprehensive product set for the HPC community.
The ever-growing complexity of scientific and engineering problems continues to pose new computational challenges. Thus, we present a novel federation model that enables end-users with the ability to aggregate heterogeneous resource scale problems. The feasibility of this federation model has been proven, in the context of the UberCloud HPC Experiment, by gathering the most comprehensive information to date on the effects of pillars on microfluid channel flow.
Large-scale, worldwide scientific initiatives rely on some cloud-based system to both coordinate efforts and manage computational efforts at peak times that cannot be contained within the combined in-house HPC resources. Last week at Google I/O, Brookhaven National Lab’s Sergey Panitkin discussed the role of the Google Compute Engine in providing computational support to ATLAS, a detector of high-energy particles at the Large Hadron Collider (LHC).
Frank Ding, engineering analysis & technical computing manager at Simpson Strong-Tie, discussed the advantages of utilizing the cloud for occasional scientific computing, identified the obstacles to doing so, and proposed workarounds to some of those obstacles.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/02/2012 | AMD | Developers today are just beginning to explore the potential of heterogeneous computing, but the potential for this new paradigm is huge. This brief article reviews how the technology might impact a range of application development areas, including client experiences and cloud-based data management. As platforms like OpenCL continue to evolve, the benefits of heterogeneous computing will become even more accessible. Use this quick article to jump-start your own thinking on heterogeneous computing.