A Toronto-based high performance computing (HPC) technology vendor is targeting “newbie users” to HPC with a pre-integrated cluster stack optimized to run on Dell hardware in an effort to alleviate deployment headaches, a company executive said.
Platform Computing Inc. released its Platform Open Cluster Stack (OCS) Dell Edition this week to tap into these new users of HPC because, according to the vice-president of product management James Pang, that’s “the growth area in the market”.
Platform OCS is optimized to run on Dell which means that while it can run on any x86 cluster, it works better on Dell hardware given the integration with Dell’s OpenManage suite of systems management applications.
The choice of Dell for this vendor-specific offering, said Pang, stems from the “much tighter OEM partnership” that existed with the company, and of Dell’s notable market share around data centre hardware.
Besides integration with Dell OpenManage, Platform OCS also integrates with third-party cluster management tools such as those from the open source community. That ability, according to Pang, reflects the array of open source technologies that are “quite good and really well-tailored to this marketplace.”
The broad integration capability essentially lets HPC installations “talk to the non-HPC side”, said Pang. Integration is essential in HPC, he said, because the primary reason for purchasing clusters is to run applications faster and cheaply, and “if they don’t have those out-of-the box application integrations, the customers just have the bare metal.”
A Web console has also been added to this release that’s “a much more visual and easier to use point and click interface”, said Pang, and is meant to help address HPC cluster management challenges, traditionally a roadblock to adoption of the technology.
The vendor-specific approach that OCS assumes, said Darin Stahl, lead analyst with London, Ont.-based Info-Tech Research Group Ltd., is likely to help with Platform’s market penetration even among heterogeneous data centres. “It’s not that you couldn’t use the platform solution on any other servers,” said Stahl, “you can but they’ve done this tight integration and an out-of-the box solution means that for deployment, you’re going to… remove a lot of inefficiency.”
Moreover, HPC requires specialized skill that is typically not core to most IT departments, continued Stahl, therefore a pre-integrated product will shortcut much of the lengthy integration time and often steep learning curve for data centres new to HPC.
Platform OCS’s capability to integrate with third-party management applications, said Stahl, will grant a “more complete view” when running clusters to ensure that application performance is evenly balanced across the board. In particular, integration with flexible open source third-party management tools, he added, will allow the necessary freedom to customize the deployment for specific applications and problems that may arise.
Platform OCS is targeted at a range of industries including industrial manufacturing, life sciences, government and research, oil and gas, and financial services.
The cluster stack is also offered in a version optimized for RedHat Linux and runs on all x86 platforms as well. “That’s another channel to market,” said Pang.