As computing requirements and the demands of end users continue to grow, affecting businesses of all sizes, the use of graphics processing units is becoming a more common practice in the industry.
More users are looking for this kind of high-performance capacity on an on-demand basis to allow them more flexibility to grow, which is a significant opportunity for web hosting companies to fulfill these needs in a cloud solution.
Recognizing this, web hosting provider SoftLayer announced earlier this week it has launched its new high-performance computing cloud-enabled solution comprised of servers powered by NVIDIA‘s graphics processing units.
The scalable nature of GPUs provides companies with greater flexibility, particularly for those organizations that are performing research, engineering and development projects.
And more importantly, providing HPC in a pay as you go model gives web hosting companies a great way to reach a new market of customers.
“Resellers who are taking a server from SoftLayer and are layering on services to that,” said SoftLayer chief scientist Nathan Day. “Those companies could have a new class of customers, where the hosting customers could obtain these servers with GPUs from SoftLayer, layer on their management, and then make those available to the end users either in the entertainment, creative, or scientific communities. This really could enable a whole new class of resellers people that specialize in this type of application and bring this type of computing power down to a market that would never have been able to obtain it before.”
The new HPC solution certainly appeals to those users in fields with compute-intensive environments, such as data mining, numerical and seismic analysis, along with advertising agencies, game developers and interactive media firms that perform video processing and 3D rendering.
GPUs work in conjunction with the server’s CPU to accelerate application and processing performance, offloading the compute-intensive portions of the application from the CPU.
As a result, SoftLayer’s GPU solutions boost application performance by up to 10-fold, provide cluster management and GPU monitoring, and allow parallel compute environment operates invisibly to the user.
“It enables a new set of workloads to move into the cloud,” said Day. “We’re not going to revolutionize the standard hosting industry, but we are going to allow those specific applications and on-premise data centers to move up into a hosted environment.”
The hybrid cloud solution features dual-processor Intel E5-2600 Sandy Bridge based servers supporting one or two NVIDIA Tesla M2090 GPUs, delivering up to 665 gigaflops of double precision floating point performance.
SoftLayer HPC servers with NVIDIA GPUs are priced from $879 per month for a configuration of one GPU, 16GB of RAM and 500GB of storage.
Day addresses the fact that there are a few competing GPU- powered hybrid cloud solutions on the market, he strongly believes that SoftLayer HPC servers provide more flexibility to customers than the other solutions.
“We admit that we are not the first people to do the GPU in the cloud,” said Day. “Amazon has a GPU offering but it’s a very specific set of hardware with no configurability. For this product, SoftLayer has done a traditional model of letting our customers self-configure the hardware to fit their needs, and still have it on a month to month, instead of trying to pick one option on a competitor’s site.”
SoftLayer’s new HPC servers are now available on-demand and can be provisioned via SoftLayer’s portal, mobile apps or API.
Talk Back: Are you currently deploying GPUs with your cloud servers. Would you say this is the general direction that the web hosting industry is moving toward? Let us know in the comments.