Rackspace officials are promising significant boosts in performance with a new line of servers the company is putting into its public cloud environment.
The company is arming its new Performance Cloud Servers with Intel's latest Xeon E5-2600 v2 processors, solid-state drives (SSDs), up to 120 gigabytes of RAM and network throughput of up to 40G bps. Rackspace is aiming the systems at such workloads as basic Web hosting to big data, and is promising better performance than what is found on the company's current cloud servers.
"These servers have been engineered to deliver exceptional levels of application performance, with greater speed, throughput, reliability and scale," Mark Interrante, senior vice president of product and engineering at Rackspace, said in a post on the company's blog. "This enables you to run workloads in the cloud, with high performance and efficiency, that could not run well—or in some cases, at all—on Standard Cloud Servers."
Rackspace officials said the servers are designed to keep up with the growing demand for enabling applications to run quickly and to scale rapidly. According to the company, the new servers offer four times the total memory of the current systems, twice the CPU performance and 132 times the total disk I/O. In addition, they offer 8.3 times the network bandwidth and 2.6 times more overall performance.
Rackspace's new systems, announced Nov. 5, come as more cloud providers are looking to grow the performance capabilities of their environments. Amazon Web Services this week began offering instances to its public cloud that leverage Nvidia's Grid GPUs to offer high levels of parallel processing capabilities for such jobs as video creation, visualization and streaming graphics-intensive applications. Organizations with high-performance computing (HPC) environments for several years have been leveraging GPU accelerators to improve the performance of their systems without driving up power consumption. Companies using public clouds are now demanding more GPU capabilities to handle parallel workloads. Nvidia's Grid GPUs comes with 1,536 parallel processing cores.
Now Rackspace is expanding the performance capabilities in its cloud offerings with the Performance Cloud Servers, which run the OpenStack open cloud orchestration technology. The first Performance Cloud Servers are found in Rackspace's data center in Northern Virginia and are available now. Later this month, the new systems will go online in facilities in Dallas, Chicago and London, and then follow in Sydney, Australia, and Hong Kong in the first half of 2014.
Commissions, which makes software for the real estate industry, has been beta testing the Performance Cloud Servers. The company, which a year ago was running its SQL Server in the public cloud, was having trouble getting enough memory and I/O speed for what it needed, so it moved the workloads from the cloud onto dedicated hardware.
According to Matthew Swanson, chief software architect for Commissions, when the company beta tested the Performance Cloud Servers with a full SQL Server Enterprise Edition, "the jump in performance was noticeable almost instantly."
It took minutes to get the database up and running, workloads ran significantly faster, and the number of transactions per second with the SQL Server went from 300 to 400 on the old systems to as many as 3,500-plus at peak and 2,500 transactions over 15-minute periods.
"Performance Cloud Servers give small, medium and large businesses the ability to operate faster," Swanson wrote in a post on the Rackspace blog site. "The SSD option is something I think all businesses can leverage at the web tier, the middle tier and at the database level. If these Performance Servers had been available a year ago, we certainly would have considered it as an option over our current dedicated environment. Now, we'll take a serious look at putting our data warehouse in Performance Cloud Servers in the future."