In a recent post on PacketPushers, blogger Carlos Cardenas considered the switch abstraction interface, switchdev...
and their roles in network design. Both SAI and switchdev are examples of hardware abstraction models, designed for application-specific integrated circuits -- otherwise known as switch silicon, or ASIC. These models serve as open source frameworks that enable ASICs to be present in software. Cardenas noted that under this framework, a Broadcom ASIC could be used the same way as an ASIC from Cavium or Mellanox.
In a helpful chart, Cardenas compared the two approaches and their respective features. Switchdev relies on a Linux Kernel, whereas SAI relies on the Open Compute Project. Switchdev allows stock Linux distribution and allows drivers upstream, but does not allow binary drivers. In Cardenas' view, developers and end users both win by seeing switchdev and SAI move forward. Developers avoid having to use "cumbersome" software development kits; end users, meantime, can use the network operating system of their choice, without waiting for vendors to retool ASICs.
"Do you need a special version of Windows to run on Intel vs. AMD CPUs? No, they conform to a well-defined interface and as such, you are free to choose which CPU you want to power your computer," Cardenas wrote. "That is what is in store for end users with open networking and the disaggregation approach that comes with it."
Read more of Cardenas' thoughts on switchdev and SAI.
Incentives to catch bugs in network software
Ethereal Mind blogger Greg Ferro recently explored the topic of debugging network software and why bugs have become an accepted part of networking software. Vendors claim they are concerned about their products, but quick development cycles are a result of customer demand, which, according to Ferro, usually comes down to speed. "Customers don't see bugs as a key issue," Ferro said.
In Ferro's view, the problem comes down to "moral hazard" and "false incentives." With IT professionals expecting bugs -- and the outages that they can cause -- Ferro said he believes there is little incentive to remedy them in the first place. Instead, technical support goes through the laborious process of identifying bugs to vendors and coming up with slow patches for problems. He pointed out the irony of enterprises hiring professional services to test vendor products for bugs in proof-of-concept trials. As a remedy, Ferro suggested borrowing a page from cybersecurity, where companies such as Bugcrowd hire freelance professionals to identify bugs for vendors before their products ship.
Learn more of Ferro's thoughts on catching bugs.
The Internet of Things ... and then what?
Earl Perkins, an analyst with Gartner, probed the future of networking after the Internet of Things. Perkins said he is no fan of the term Internet of Things, which he believes is derived from marketing literature and doesn't accurately reflect the technology, which includes many devices connected to private networks, rather than the Internet.
Given the presence of interconnected digital devices -- everywhere, from our bodies to the manufacturing floor and from medical devices to deep space --Perkins suggested that we are now witnessing the beginning of what he called pervasive digital presence. PDP goes beyond operational technology (OT), he said. Rather, PDP reflects ubiquitous and standardized interconnectivity via ever-cheaper devices. To embrace a future defined by PDP, Perkins recommended a framework that includes secure development of applications, particularly for machine-to-machine communications, the integration of IT and OT security, and the adoption of foundational security hygiene.
Explore more of Perkins' thoughts on the PDP.
Microsoft OS and SAI
The rise of bug bounties
IoT's murky future