The need to get content into users’ hands, around the globe, in seconds has necessitated the growth of content delivery networks (CDNs) like Akamai, CloudFlare and Fastly. CDNs are becoming a core component of many large-scale systems, so it’s only natural that their usage would expand beyond simply caching files that are in high demand.
Fastly Chief Technology Officer and co-founder, Tyler McMullen, is thinking hard about what this edge computing scenario might look like. As we chatted with him in this latest edition of The New Stack Makers Podcast, he ruminated on the possibilities and even the definition of edge computing, as a whole.
To some, it would mean doing computation anywhere but the core systems layer: in the pipes, at the CDN or in the end user’s device. That last bit even implies the use of edge computing in IoT devices. The idea is appealing, as it would save round-trips and server load, but in practice, it does add another layer of complexity to an already tricky stack of distributed content delivery.
McMullen said that the big question is just what exactly is edge computing? “I get asked this question a lot, and the problem is depending on who you ask … If you ask a mobile developer of some kind, the edge to them is the device itself … I’ve heard stories of [telecom] companies that are thinking about putting tiny servers along the sides of highways. So, the edge to them, is essentially the edge of their network.”
He continued, “What you consider to be the edge seems to depend entirely on your perspective within the network. It seems to me when we’re talking about the edge, what we’re talking about is anything that really comes between the user and the application’s origin.”
McMullen sees edge computing as an enabler of better service offerings, through the shortened response times available when individual nodes can make decisions for themselves. “We provide a way for people to write code that is spread around to thousands of servers at the same time, spread as close to your users as you really can get without being on the devices themselves.”
McMullen has lots of other thoughts on edge computing and on the modern distributed computing ecosystem. Oh, and he’s also into Lisp! Be sure to check out the whole podcast above.
- 1:14: What does edge computing mean?
- 4:14: What are people actually doing in the wild with edge computing?
- 6:00: How do you coordinate actions within a system with so many decision makers?
- 9:02: Who is doing interesting academic research around some of the problems associated with edge computing?
- 10:19: What does this look like for you and for the future?