Facebook's Open Source Project (OCP) is hoping to ramp up the rate of innovation and deployments for service providers with its concepts of open source computing.
After starting out five years ago as a small electronics team in the basement of Facebook's Palo Alto, Calif. headquarters, OCP has simplified and modernized data center technology with the disaggregation of networking hardware and software. OCP has gained traction in the data center sector by kick-starting open source contributions for networking, servers and storage.
In January, OCP announced its OCP Telco Project, which was designed to bring those same open source computing technologies into the realm of telcos. (See Telcos Board OCP Bandwagon.)
The charter members of OCP's telecommunications forum included AT&T, Deutsche Telekom, EE, SK Telecom and Verizon.
In a room packed with attendees, the Open Compute Project hosted its OCP Telco Engineering Workshop at Light Reading's Big Communications Event in Austin in May. To find out more about OCP and its OCP Telco Project, Telco Transformation interviewed Omar Baldonado, OCP co-lead and manager, network team at Facebook.
Telco Transformation: How are OCP developments going to impact on telco operations and skills/recruitment strategies?
Omar Baldonado: OCP, as well as most of the IT industry, is focused on disaggregation: the breaking down of traditional data center technologies into their core components to build new systems that are more flexible, scalable and efficient. This allows best-of-breed and open providers to contribute different hardware and software components.
Telco operations, architecture, and engineering need to become comfortable with an integration mindset and recruit those who have the skill set to do so -- because for telcos to integrate OCP components, it'll be important for them to understand the constituent components. OCP networking is particularly focused on the disaggregation of networking hardware and software, so having software expertise on the team will become more and more critical over time.
TT: Adopting OCP technology will require a new mindset, an awareness of new "languages" related to open source etc. that folks in telcos may not have. What are those new skills and IT languages?
OB: We sometimes say that OCP is bringing open source to hardware. Open source brings many benefits: flexibility, extensibility, and a community of people to work with -- but it requires the integration mindset and skill set mentioned above. Further, if companies are interested in actual open source software, then they'll need to hire software engineers.
TT: Is there a cultural transformation for telcos that also needs to take place in order to leverage OCP technologies?
OB: Organizations within telcos need to support the development and time that it takes to build teams and projects around disaggregated networks. Culturally, telcos need to embrace disaggregation at all levels of the organization and be prepared for a learning curve. This experience can be compared to the past shift many enterprises and web companies made to Linux. For a while, Linux deployments were very problematic and had new and unique challenges. Thankfully, some companies made the cultural shift to adopt Linux no matter what, and it allowed them to ride out the bumps and come through on the other side in a better place -- benefiting from the new technologies.
TT: What impact might the adoption of OCP technology have on telco data center strategies? Could this tip more towards developing their own, or go the other way?
OB: Through OCP, companies share not only the design of individual components -- like servers, switches, and storage devices -- but also a guide as to how the components are plugged together to make a rack, pod, or even an entire data center. The OCP ecosystem has many examples of companies learning from each other in an open way. The most recent example being Yahoo Japan, who built its big data Hadoop data center in less than a year after speaking to OCP members like Facebook.
I believe telcos can look at the existing data center designs within OCP and modify them to suit their particular needs. This is the benefit of open source; you can base your work on publicly available materials and knowledge, but still have the flexibility to change it to suit your unique needs.
TT: What does OCP mean for network management teams?
OB: Facebook believes in treating network devices more like servers, especially from a management perspective. Our open contributions of software, like FBOSS, and hardware, like Wedge and Wedge 100, represent this, and we actively use many server management packages for our network devices. By treating switches more like servers, companies can leverage the provisioning, automation and monitoring infrastructure that are available for servers.
For example, at Facebook, billions of alerts come monthly from networking devices. The Facebook network engineers wrote software that automatically processes those alerts, discarding the vast majority and automatically taking remediation action on those alerts that need it.
In addition, the OCP networking switches have commercial and open source software options that present similar network management interfaces as non-OCP networking devices. So, the standard network management practices can be maintained.
TT: What are your final thoughts about telcos working with OCP?
OB: There are a lot of opportunities to apply contributions and learnings from the existing OCP networking, server and rack projects to telcos. Telcos are in a unique space as they deal with server clusters that range from 1-2 racks near the edge to larger compute clusters with hundreds of racks. I also believe there will be a lot of work around new software, especially with service deployment and management for telcos being based on workflows that are software-automated. The industry continues to move towards faster and better service deployment.
— Mike Robuck, Editor, Telco Transformation