![](https://d15shllkswkct0.cloudfront.net/wp-content/blogs.dir/1/files/2015/08/PLACEHOLDER-SiliconANGLE.png)
![](https://d15shllkswkct0.cloudfront.net/wp-content/blogs.dir/1/files/2015/08/PLACEHOLDER-SiliconANGLE.png)
Open sourcing hardware and software is “core to [Facebook, Inc.’s] mission,” according to Matt Corddry, Director of Hardware Engineering at Facebook. In a live interview with theCUBE co-hosts John Furrier and Jeff Kelly during OCP Summit 2015, Corddry explained that the benefit to the internet behemoth, he explained, is that it “actually helps us move faster.” By open sourcing hardware, Corddry added, Facebook intends to “connect more folks,” and get more smart minds on the problem.”
Part of Facebook’s recent software innovations “blur the boundaries of open hardware and open software,” Corddry said. He called out FBOSS and Open VMC as particularly exciting products — software products that Facebook is announcing at a hardware summit precisely because they bridge hardware and software.”FBOSS allows you to program you won network switches,” enabling folks to “hack [their] own software on [their] own network switch.” Open VMC, he said, “allows you to roll your own code in the little baseboard management controller Both these innovations are uncharted territory, he explained, because they enable “people to hack and innovate down at the hardware platform management level.”
While Facebook’s announcements at the Open Compute Project Summit leverage bridge hardware and software, the company also has invested in innovating in both areas separately.
Recent software innovations, like the HipHop Virtual Machine, are designed to make it easier for Facebook to “connect the world.” The HipHop Virtual Machine allows “people to scale out large web properties” thanks to “much more efficient PHP execution.”
Corddry also highlighted the System on a Chip design that Facebook recently announced. “A lot of it is the disaggregation of our infrastructure,” he said, adding: “Instead of cramming all bits and resources into one big box…we’re going to really focus on solving one problem at a time.” The intention, Corddry explained is to “scale out massive amounts of compute without needing to put a bunch of local storage and other resources in the box at the same time.”
Hardware-based innovations are in Corddry’s wheelhouse. “My team,” he explained, “mostly focuses on the gear in the data center.” Contrary to popular trends, Facebook is building “big, ugly tin boxes” instead of going “smaller, smaller, smaller” servers. The benefit of theses servers, said Corddry, is that they’re “super efficient,” and “built to work at massive scale.” In particular, Corddry highlighted that the amount of power required to cool one of their designs is only three to six watts, whereas traditional OEM designs “take 80 watts of power.” That efficiency, he commented, “makes a tremendous difference.”
Not every company is equipped to experiment and build from scratch the same way Facebook can. “It makes sense at our scale,” said Corddry, “but if you’re a tenth of our scale, it wouldn’t make sense to make that investment.” Open compute, he explained, is a great solution to overcoming this obstacle. It enables companies to “work with companies [they] know and trust.” Calling out Apache Hive and Hewlett-Packard Development Company, L.P. in particular, Corddry said they “bring efficiency wins, scale out wins, and open modular approaches” into client’s data centers.
Furthermore, Corddry highlighted that open compute has lifted the veil from enterprise deployment’s edges. Now, those departments realize, ” we don’t need to buy the really high end stuff because a lot of our problems translate really well to open compute hardware.” Corddry stated. He added that these same folks have a new perspective on “the myth of hardware reliability.” It has become clear now, Corddry explained, that “the failure rate for open compute hardware is better than what we saw in white box, OEM hardware in the past.” Revelations like these, he said, are why Facebook is dedicated to “creating a more reliable system through its simplicity — it’s such a simple design, there’s just nothing tin there to fail.”
Watch the full interview below, and be sure to check out more of theCUBE’s coverage of OCP Summit 2015.
THANK YOU