Credit: Teerawut Punsorn
Much of the discussion about the Open Compute Project has centered on its impact on Facebook -- that is, how much money and energy it saved -- or on the contributions of the directly participating companies (such as Microsoft).
But as more hardware makers start offering OCP-inspired designs, it's worth reflecting on how that might influence what and how manufacturers produce hardware for the data center. Here are four possible ways the OCP might influence what they do.
OCP effect No. 1: Faster iteration between designs
Product cycles originate with the creator, not the consumer. For the data center, this means change is locked all the more closely to the whims of the hardware makers, rather than the folks who assemble and use the data center.
With OCP designs, though, the iteration process isn't locked to any one manufacturer. If a brilliant new idea appears in data center design, it theoretically becomes easier to put it into play and see how it holds up under real-world loads -- not just because of OCP hardware, but also due to OCP design-sharing methodologies.
This may be doubly important for the promised wave of ARM-based server designs (whether from Microsoft or others) set to be added to the OCP hardware arsenal. The more standardized the overall design, the easier it ought to be to offer new iterations of the system-on-chip designs found at the heart of such servers.
OCP effect No. 2: Consistency of designs
Consider this the flipside of the first point. The OCP's reference designs make it theoretically easier for a data center's hardware to be consistent and to avoid the arbitrary discontinuation or gratuitous changes that can creep into a given vendor's offerings. (For example: Why did my blade suppliers suddenly change to that vendor of storage controller?)