[gorkish]

I think there is a severe miscommunication of what is meant by "edge"

To me, a technical person, the "edge" is the work-at-home office, a retail store, a 5 person branch, a jobsite, a delivery van. If "edge" was really happening at the scale that marketing departments want you to believe, there would be better hardware available -- short depth servers, cluster-in-a-box hardware, etc. I don't see this stuff really coming out in any quantity outside of some niche stuff that is clearly targeting fortune-1000-type large businesses.

I've been around a while, and I think I finally figured out that all of the above applications are just the carrot on the stick. For the everyday business, "edge" just means "server room." All the companies who were stupid enough to remove all on-prem computing not so long ago now need to buy it back, but the MBA's need a new word to avoid the untenable position of having made a mistake. Somehow it reminds me of a cat burying its own shit.

We still want legitimate edge compute hardware, btw!

[walolla]

Hey, thanks for the response!

Just to clarify, by "edge" I mean the combination of location and runtime of the serverless functions. Basically a bunch of places where your application is deployed automatically with a smaller runtime. Just like classic serverless functions (from AWS for example), but more optimized, which kind of fixes some issues like cold starts. Please note, that I'm more of an outside observer, so I could be missing a lot, that's why I'm asking more experienced people :).

Of course, it's not a "new industry standard" kind of thing, but I'm sure that Amazon and Cloudflare (and mainly hardware companies, once they notice it) will come up with something to optimize the hell out of this new approach.

The point is exactly that it's much easier to pay $5 a month to make your app work fast across the world after deploying it with 3 clicks, rather than setting up a box, or god forbid a physical machine (now you have to deal with outages, networking problems and scaling). I mean, it's good to remove bottlenecks from the process of creating new things, but as I said, something still bugs me about it (maybe the complexity, the level of abstractions, etc.).

Of course, thanks for your thoughts!

[kube-system]

Yeah, there are definitely different interpretations of 'edge' by market. I think, for smaller businesses, there's not really a need for heavy computing at the edge. For them, traditional PCs or tower servers can handle the load just fine.

And any business that is large enough (e.g. 'fortune-1000-type' or anything close to it) to need heavy compute at the edge also can afford a network closet with a 42u in it.

I just think there's not really much of a market between "I need more than this PC can do" and "We don't have a network closet"