8 ms·
For everyone slating this article, please try thinking of it from a different perspective (and look beyond the buzzwords). Here's my take. (paraphrasing) 'In t
by amirmc 12y ago
For everyone slating this article, please try thinking of it from a different perspective (and look beyond the buzzwords). Here's my take.
(paraphrasing) 'In the rush to 'the cloud' people are disovering that there are bottlenecks related to bandwidth to edge devices (phones etc) and this problem will only increase as more and more devices become conencted (cf Interent of everything). However, not all services/products need to constantly push data back and forth to a centralised place. There is enough processing power in many devices such that more processing can happen at the edge so why not build more applications this way? Cisco and IBM see merit in this approach and hope to capture part of a (presumabley nacscent) market by providing some kind of processing ability that is a little nearer the edge of the network, rather than huge datacentres.'
Is that any better?
My view on the article is that until developers can more easily build distributed applications, then we're stuck with the prevailing paradigm of large centralised services. I don't see how anything from Cisco or IBM changes anything other than having a slew of mini-datacentres that essentially do the same thing.
What we really need are the means to create, deploy and maintain software in a more distributed manner. Having all the fancy smart sensors in my home talking to my hub is preferable to having them all talking to their individual silos. At least then, I can excersice some control of what's going on within my network and benefit from e.g. my high-speed wireless vs my much slower upstream bandwidth.
FWIW I'm working on an open source toolstack to make building distributed systems a lot more stratightforward. Essentially, we want to get to a place where everyone can own their own piece of the cloud. Whether that ends up being called the 'fog' or 'dust cloud' or 'motes' is beside the point. You can read an overview at http://nymote.org/blog/2013/introducing-nymote/ http://nymote.org/blog/2013/introducing-nymote/
- rsync 12y ago"There is enough processing power in many devices such that more processing can happen at the edge so why not build more applications this way?" ... and thus spins the endless wheel of IT. You could find this exact same sentence in any number of trade journals going back every five years to 1970. Just replace "mainframe" with "cloud" and "client server model" with "fog". See you in another five years...
- amirmc 12y agoI don't understand what you're getting at. That things go in cycles? Sure, I agree with that. Is there something else?
- ghkbrew 12y agoHe's talking about a specific cycle in computing where it has moved from centralized processing to decentralized multiple times in the past. For example -first there were mainframes with glorified type-writes as terminals -then we had personal computers -then we had thin clients -then we had thick clients -etc Basically this has happened before, will likely happen again, yet people (or maybe just tech reporters) always seem so surprised when the pendulum swings back the other way
- jlarocco 12y agoI'm not sure anybody is surprised here. The fact that the pendulum is swinging back the other way is news, though maybe not very exciting news.
- mark_l_watson 12y agoThanks, that is a bit better. I would hope for more applications that use cloud service to hook up apps on mobile devices, that would then directly peer to peer. I have fairly broad experience in networked applications since the 1980s, but this is one application area that I have never got my fingers on except for peer to peer UDP connections in a few networked games (LAN only). It will be really interesting to see the future effect of IPv6, more use of lossy but more efficient broadcasting protocols, etc.
- amirmc 12y agoI'd be really interested in your thoughts on one of the things we're building. It does exactly what you describe. http://nymote.org/docs/2013-foci-signposts.pdf http://nymote.org/docs/2013-foci-signposts.pdf
- mark_l_watson 12y agoThanks for the linked PDF. I skimmed it and bookmarked it for when I get home from vacation. I got your contact info from your profile and I'll contact you when I am home.