As a sysadmin, any time I read some variation of; "At some point even data centers may become a thing of the past.", I know they don't know what they are talking about. As a matter of fact it has provided much joy through laughter (followed by required sysadmin scotch) at the show "Silicon Valley" for obviously parodying the issue. Datacenters aren't going anywhere, and this strange fascination in hipster-hackers with instant uber-decentralization-pushes concerns me because it ignores some of the more real (and fixable) issues at hand like dns centralization in favor of magical "p2p(+blockchain) will save us all" thinking not backed by much real world practical implementation.<p>Don't get me wrong, I'm a darknet, meshnet supporter. I love decentralization. That said, I support the establishment of the infrastructure required to support it independent of end-user devices, and I think for security and other purposes it's at least possible they should remain separate, and devs shouldn't assume so much right to cpu-cycles.<p>So in essence the topology I think that is preferable would be properly called decentralized-distributed.<p>Of course thats part of the reason I support things that go against that common grain, such as ipv6nat.
I wish one of these "next-generation technology" blog articles would mention Named Data Networking[1]. I think the real problem with the web is the IP protocol itself, not any higher-level protocol or particular way of using IP.<p>From a very high level: The Web should be built on top of a broadcast protocol, not a point-to-point protocol. If it were, many complex issues would be solved.<p>[1] <a href="https://en.wikipedia.org/wiki/Named_data_networking" rel="nofollow">https://en.wikipedia.org/wiki/Named_data_networking</a>
So there are a number of issues that are not really answered here.<p>1) the cloud is a costed model. you pay a company to look after your stuff, so you dont have to manage it yourself<p>2) the cloud _is_ more expensive, but not compared to hiring your own infrastructure people (depends on scale of course)<p>3) distrubuted networks of things are much much slower than centralised. save for a few exceptions.<p>4) Trust. even though you stuff should be encrypted, spaffing it out in the open is a large risk, because should your old keys leak, it'll be trivial to retrieve old data.<p>5) money. who on earth is going to buy, run and maintain this infrastructure? whats in it for them? Who organises development of tools, patching etc? who do I call when there is an outage.<p>in short for this to work, it would required a 180 degree culture shift of capitailsm.<p>also physics, to make globally distrubuted systems both fast, relaiable and consistent
IMHO "cloud" as a whole was way overhyped for the value it provided. While it delivered on simple requirements any thing complex meant things got out of hand quickly. Quite a lot of applications were sold as "to be used by business user" but then setup ensured the complexity was so high that it required whole dedicated technical teams to manage it.<p>That said, is it the end? It is doubtful - A lot of conventional, old school companies which were against moving to cloud because of various reasons are now seriously considering cloud. Quite a lot of these companies are heavily dependent on Oracle, SAP etc which is just now rolling out/pushing their cloud products. Maybe it is the shoe shine boy and Joe Kennedy moment, but it is difficult to tell.
>This is a problem mostly because of the way we’ve organized the web. There are many clients that want to get content and use programs and only a relatively few servers that have those programs and content. When someone posts a funny picture of a cat on Slack, even though I’m sitting next to 20 other people who want to look at that same picture, we all have to download it from the server where it’s hosted, and the server needs to send it 20 times.<p>The article reminds me of the Pied Piper platform from <i>Silicon Valley</i>, but especially the above quote.
Why does the author implicitly assume that the amount of spare local storage is enough for the swarm+redundancy? Furthermore how many backups are required for data to ensure the same level of protection as a modern cloud provider? Does the math even work?
While it is _technically_ possible to build heavily decentralized, managed hardware deployments, no one has figured out yet how to charge for them except in the case of very large customers. Until this happens, there will be no swing back to anything decentralized.<p>Although I do understand the allure: just about anyone today could very economically purchase 5-10 servers with 10-18 cores each (and those will be _real_ cores, not hyperthreads). There's nothing impossible about automating software management on such a thing, even to the extent that you'd get in cloud (VMs, containers, distributed storage, automated updates, VM migration, network partitioning, etc). I believe Microsoft will lease you a fully set up shipping container with Azure in it, all that needs to be done is scale this down.<p>But again, how does one extract billions of dollars in profits from something like that?
I never understood the difference between Mainframe Computers / Computing and Cloud Computers / Computing.<p>In both cases, a centralized server connected to the internet does all your application, storage and processing. If that server goes down / connection is lost, in both cases you and your "SaaS" applications are SOL ( Sh*t Out of Luck). Also in both cases, you offer hackers a centralized location to target and hack.<p>What's the difference, except in the name and in the marketing ?
At the core of the cloud paradigm and the centralized/decentralized issue lies no technical problems but a political one about control and privacy invasion.<p>The only reason why we have youtube instead of VLC + eDonkey packages are because of lawsuits. It is a technological solution from 15 years ago that is technologically superior and much harder to censor.<p>Datacenters and "cloud-based" services do not answer to a technical problem but to a political one: how do we control information flow in a decentralized net? The answer: provide bottleneck of information for free.
Even if all the arguments are true and take into account all tradeoffs -- which I don't think is the case -- there is a big difference between "will eventually happen" and "is coming" or "we're facing the end".
> The compute speed and storage capacity of computers are growing faster than the bandwidth of the networks.<p>So your proposed solution is to spread out compute and storage to hosts that are even more network constrained???
The article also assume that the client bandwidth is unlimited, but in reality more and more ISPs have data caps.<p>If a p2p network would become this large, the need for more client bandwidth would increase exponentially.
As I’ve written before, privacy and risk mitigation are the two reasons this article is wrong. The cloud is already the pinnacle of decentralization, and distributed web is just a fad.<p><a href="https://nrrd.io/privacy-and-risk-mitigation-are-the-biggest-threats-to-the-decentralized-web-b685ce7cf125" rel="nofollow">https://nrrd.io/privacy-and-risk-mitigation-are-the-biggest-...</a>