22 ms·

The battle of the computing clouds is intensifying

178 points by acqbu 5y ago

CraigJPerry 5y ago
https://archive.md/meT9E https://archive.md/meT9E
asimpletune 5y ago
I wonder when there will be cloud co-ops?
- pjkundert 5y ago
  Now: https://holo.host https://holo.host Ready for testing and pre-release R&D.
  - vgel 5y ago
    IMO, an ideal co-op cloud host wouldn't have a bunch of complicated p2p junk. Just a good ol' datacenter and people to operate it, with the cost and governance shared by the organizations running workloads on it.
  - 4ggr0 5y ago
    What's your connection to Holo? You seem to mention them a lot.
    - pjkundert 5y ago
      Yes, I do; sorry about that! They seem relevant, a lot... I was on the development team for a couple years; now, I'm independently developing some technology on the platform. I'll announce it soon, I hope. With the recent AWS outages, and systematic de-platforming of some legal businesses, there seems to be a growing realization that dependence on centralized hosting platforms is ... dangerous. Holo / Holochain seems to provide a unique solution to some of these problems.
      - jatone 5y ago
        pure FUD. bugs happen decentralized systems can be even more problematic to resolve issues in due to their decentralized nature and the usual requirement that fixes be deployed system wide.
      - 4ggr0 5y ago
        Sounds interesting, and thanks for disclosing your connection to Holo! :)
- helpfulmandrill 5y ago
  Would be much better than getting locked in to a company you have no control over.
- praptak 5y ago
  The other direction is to use the commercial clouds but aggressively defending their status as a commodity. Have VMs at multiple providers and move your load between them as needed.
- streamofdigits 5y ago
  This. Turn the "cloud is somebody else's computer" on its head and own part that computer. Why not "own" AWS, GCP, Azure etc.? Because you can't. They are part of conglomerates that come with a moral baggage that is too heavy for some. ps. No "hyperscaling" ability required. The idea that business activity is either complete dominance or death is part of that unfit-for-purpose mentality.
  - Layke1123 5y ago
    I do actually own shares in some of those companies. That's kind of the point. You can own the internet, and that's before even "Web 3".
- rambambram 5y ago
  Cool idea, then we can call it 'the internet' again.
  - streamofdigits 5y ago
    its not exactly the same. there are probably significant efficiencies (energy, security etc) in provisioning and sharing "cloud" hardware in reasonably large chunks.
    - rambambram 5y ago
      You are right, but I saw some striking similarities, and needed to joke about it.
k__ 5y ago
Akashnet is an interesting alternative. Sadly, it's Docker based.
- imglorp 5y ago
  Would you elaborate on the downside? That seems like a common subset that would satisfy a lot of use cases with minimal ceremony.
  - k__ 5y ago
    Yes. It's good enough, I guess. I think, a VM approach would have been a bit more flexible.
  - itisit 5y ago
    > Akash Network leverages 85% of underutilized cloud capacity in 8.4 million global data centers, enabling anyone to buy and sell cloud computing. [0] How do you even begin to operate a consistent security model across "8.4 million global data centers"? > Auditors on the Akash Network review cloud providers and digitally sign the provider on-chain with their certificate. If you only accept bids from audited providers this means you are trusting the Auditor/Provider not just a Provider. [1] For a business at scale, I don't see this being solid contractual ground. Now I have to trust Auditors to continuously vet Providers? Simply too tenuous for my comfort. [0] https://akash.network/about#vision https://akash.network/about#vision [1] https://docs.akash.network/glossary/security https://docs.akash.network/glossary/security
1cvmask 5y ago
An article that regurgitates Gartner talking points is a tainted one to begin with. Gartner even famously said that Apple should quit the hardware business: https://www.zdnet.com/article/why-does-the-it-industry-continue-to-listen-to-gartner/ https://www.zdnet.com/article/why-does-the-it-industry-conti... https://www.horsesforsources.com/gartner_fail_automation-AI_080418 https://www.horsesforsources.com/gartner_fail_automation-AI_...
- klausjensen 5y ago
  Twice in my career I have been involved in evaluating enterprise products for a large corporation - and part of that process was looking at Gartner reports. The Gartner reports were complete and utter garbage, filled with factual errors. Looks like they were written by interns, regurgitating marketing material - completely removed from reality. Since then, whenever I see somebody mention a Gartner report, I immediately dismiss that part of the argument.
  - bsenftner 5y ago
    Can confirm. Was a developer for 20 years, got an MBA 20 years ago and have been dismayed by the quality of Gartner reports. Worthless for anyone with experience, and dangerous when used to guide the non-technologist. (Yes, I've been developing for 40+ years.)
    - greedo 5y ago
      But mgmt LOVES Gartner reports. It helps them avoid/shift blame when a purchasing decision goes awry. The first question our VPs ask about a new technology is "Is this %VENDOR% in the Magic Quadrant?"
  - TheRealDunkirk 5y ago
    I am convinced that Microsoft underwrote them to boost Windows and cast FUD at everything else in the 90's, and they've never recovered their disingenuous ways. What annoys me most are the market share reports, implying that Microsoft owns the entire computing world. They don't. For personal use, I'd say Apple has broken into the majority. I'd pay money out of my own pocket to see PC vs Mac sales numbers with corporate purchases removed, rather than the same report they've published every quarter for 20 years now.
revolvingocelot 5y ago
>Ms Wang and Mr Casado have suggested that firms should think about building their own private clouds to keep costs down Yeah I distantly remember "on prem", too. Does everyone, everywhere really need to be equipped to hyperscale at a moment's notice? Wouldn't it be better not to be beholden to a megacorp that serves you with 30 pages of opaque billing for business-critcal functionality? Doesn't hiring external consultants to manage your fucking cloud infrastructure seem like sort of a risk as compared to keeping the services-wrangler in-house? ...I don't have an MBA, so these are half-serious questions. If anyone can chime in with (at least) a half-serious answer, I'd halfway appreciate it. I understand that The Economist is the voice of global capital, and global capital has decided that there's money in them thar clouds, but still!
- arcbyte 5y ago
  It's not about hyperscale, it's about time to market. Traditional corporate datacenters were dogshit when it came to provisioning new resources for business capabilities. In today's world, we need the ability to spin up and down deployments very quickly. We don't need huge numbers of resources, just fast response.
  - benjaminwootton 5y ago
    I have worked with hundreds of companies moving to AWS, and believe that this is the answer. People are investing in cloud because its faster, more agile and better, not because it's cheaper. The top line looks expensive, but if you then do the correct analysis on the cost side of the equation (taking out staff costs and DC costs) then I think it is either close or in the black from a cost perspective. If the business case was really bad, I don't think we would see the rapid uptake we have seen.
    - hectord 5y ago
      If you think from an economic standpoint, Amazon should price AWS only a bit cheaper than the on-prem options. It's a way to maximize their profits. Then you have competition (GCP, Azure, Alibaba) which will tend to lower the price, but for now the cloud industry grows. Their goal is to attract new customers who didn't use the cloud before.
username223 5y ago
Problem: Managing server hardware takes money and people. Solution: Subscribe to a middle-man that manages it for you. Problem: The middle-man is slowly turning the screws to increase its profits. Solution: Subscribe to another middle-man to figure out how to decrease the first middle-man's profits. It seems to be a universal truth that computing people try to solve all problems with another level of indirection.
- yobbo 5y ago
  > Problem: Managing server hardware takes money and people. For most, this was never the effective reason. Most rationales behind cloud migrations are likely based on quick scaling, which is meaningless in most cases - except for the signalling of committing to a future where demand hockey-sticks upwards. It's valid if you are netflix, but how many netflixes exist in the world?
  - jatone 5y ago
    its not even valid for netflix anymore most likely.
- pm90 5y ago
  > It seems to be a universal truth that computing people try to solve all problems with another level of indirection. This isn’t about computing but about specialization. Should McDonalds be in the business of operating Data Centers? Or should they just buy it from those who do it best and focus on selling food?
  - baq 5y ago
    i don't know. should a book store be?
    - pm90 5y ago
      It should not.
  - username223 5y ago
jrochkind1 5y ago
> Mr Garman counters that the higher price of moving data off a cloud (“egress” in the jargon) reflects the higher costs of that exercise. Almost by definition, customers leave with more data than they entered with. Uh, they charge more per byte, that's not what's going on. I don't know if that was AWS's Garman being deliberately misleading, or the Economist journalist misunderstanding.
jasode 5y ago
>Businesses’ main motive for moving to the cloud was never about cost but “scalability”: having access to additional computing resources with a few clicks. That's not accurate because many businesses do migrate for cost reasons. Especially for internal non-public facing servers like email. E.g. They moved from self-hosted Microsoft Exchange servers to cloud-managed email like Office 365 Exchange Online -- or switch to another cloud competitor like G Suite. >Ms Wang and Mr Casado have suggested that firms should think about building their own private clouds to keep costs down. So far few firms have opted for such “repatriation”, which is both costly and makes it harder for businesses to enjoy the benefits of essentially unlimited computing resources in the public cloud. As a case study, The Guardian newspaper tried building their own private cloud and failed. They ultimately just went with AWS: https://web.archive.org/web/20160319022029/https://www.computerworlduk.com/cloud-computing/guardian-goes-all-in-on-aws-public-cloud-after-openstack-disaster-3629790/ https://web.archive.org/web/20160319022029/https://www.compu... If the company's business profit is not in the tech sector (like The Guardian), this means its IT datacenter operations is not its core competency which makes it hard to build internal IT that can keep up with AWS capabilities. Tech businesses like Dropbox, Facebook, can hire & maintain skilled IT staff like AWS but most other non-tech businesses can't.
- rmbyrro 5y ago
  They say the "main motive", not the sole motive. Cost is also a motive, for sure, but do you consider they'd move to reduce costs if there wasn't scalability?
- krab 5y ago
  > The Guardian newspaper tried building their own private cloud and failed What I see often is that it's hard and costly to replicate the whole cloud experience in-house. OpenStack is a behemoth to operate. However, if you build something more specific to your needs (and often mix in selected public cloud services), you'll get a much cheaper stack. Small businesses can often run very efficiently with a few rented physical servers in a colo, G Suite and AWS S3. Larger ones can provide a platform specific to the technologies used in-house and combine with public cloud for any other tech stacks.
eole666 5y ago
> The situation is analogous to a century ago, when electricity became an essential input You can do nearly everything "the computing cloud" offers without it : just use servers, install your services on it, manage it, invest in more servers/ more powerfull servers if needed... Cloud computing is not a brand new invention, it's just delegating the server and services management to others... Whereas, a century ago, electricity was actually an essential input.
- londons_explore 5y ago
  In the future, I can imagine cloud providers offering discounted bandwidth from one AWS customer to another. Then I can imagine SAAS companies offering discounts if your API requests originate from AWS cloud. Maybe even Amazon prime members will get free bandwidth from their home internet connections into and out of AWS. At some point, the whole lot becomes a walled garden, and being outside it is more expensive if you need to use services that are inside.
  - rmbyrro 5y ago
    Cost of SaaS or API prices are usually orders of magnitude higher than their infrastructure marginal costs. Although this idea makes practical sense, from a financial standpoint I don't think there will be enough incentive. Bandwidth savings in infra will represent <1% savings relative to the SaaS/API price.
- martincmartin 5y ago
  Companies used to run their own electrical generators. That seems analogous to running your own servers.
  - mschuster91 5y ago
    > Companies used to run their own electrical generators. ... and have the staff with experience and resources to keep them running. These days, UPS maintenance is sourced out to the cheapest bidder and testing is infrequent, with the predictable result of shit hitting the fan at the exactly worst time.
robertlagrant 5y ago
Either The Economist or I don't know what Snowflake is.
- DangitBobby 5y ago
  I had that same thought at first, but I think they were trying to do a tie in for enterprise SaaS. They mention Hashicorp as well.
exdsq 5y ago
My first job was IT support for on-prem systems and, as I battle my containerized deployment scripts this morning, really miss those days. Probably just nostalgia but I had named them all after superheroes and it was just way more fun when you got to spec a server and install it, fix them (physically) when they had issues, and panic when the AC broke. Edit: Also I was way more fit from climbing around a building all day cabling stuff. Genuine question - does anyone know what sort of IT jobs require physical work? I miss those days a lot!
- VTimofeenko 5y ago
  > does anyone know what sort of IT jobs require physical work? I have seen a few positions in the US for datacenter technicians. Those positions had explicit requirements(probably because of ADA) to be able to lift over X lbs as part of day to day activities.
- marvinblum 5y ago
  Hetzner would probably be the perfect working space for you. They build all of their servers by hand on custom racks. You can move around in 14+ buildings in Falkenstein, manage cables, fix servers, work on their cloud offering, ...
  - teloli 5y ago
    AND you would get to live in Falkenstein, one of the most exciting places in Europe.
    - mellavora 5y ago
      which does indeed sound like a good home for servers named after superheros.
- imadethis 5y ago
  COVID has probably changed this, but networking jobs tend to require a lot of racking and cabling work, especially if you’re fitting out a new work area. Even with the majority of your infrastructure moved to the cloud, you’ll still need APs or Ethernet jacks wired into a LAN for the end users.
jamesliudotcc 5y ago
Also, security. Has anyone noticed that all of the big ransomware cases have been for on-prem servers, not cloud?
- betaby 5y ago
  Not really. On other side news like 'my AWS account racked 40000$ while I was asleep (because of the crypto-miners)' make to the front page way to often.
nderjung 5y ago
One way to reduce compute costs significantly is by using a much lighter stack. Typical deployments to the cloud are based on the traditional OS model, and more often than not are a container-in-a-VM solution to deploy some service. The reality here is that there is so much unused, unnecessary and wasted compute resources. Why deploy a full OS and kernel which will only ever serve static web content, for example. And then there's the attack surface that comes with a full OS, running additional services like SSH which may only ever be used once. Major vendors like Amazon and Google are happy to provide this to anyone and with simple on-boarding, but, as is clear in the article, it's expensive. With Unikraft[0], you can completely re-package your application as a unikernel, automatically. It's a single VM image and does nothing more than just running the application. It boots straight to the main application, it doesn't have a shell, and has nothing else running along side it. The result is a huge cost saving (up to 50%), since you're no longer having to pay for these extra resources just to run the a full traditional OS image plus your application. [0]: https://unikraft.io https://unikraft.io
- mikkelam 5y ago
  Maybe you should also mention that you are affiliated with this project. This reads like an ad.
  - nderjung 5y ago
    Whoops :) Yes I am affiliated, I can't update my comment but I'll make sure to indicate this if/when I write a comment which discusses anything about Unikraft.
- pid-1 5y ago
  50% compared to what? AWS EC2? Lambda? Google Cloud Run? Those sort of numbers without even having a pricing page give me 0 confidence sorry.
  - nderjung 5y ago
    We are launching a pricing page in the next month as we are gearing up for launch :) One example of recieving a 50% cost saving is from Amazon Machine Images (AMIs). We compared the cheapest official NGINX AMI available on the Amazon Marketplace[0] and then also ran an NGINX+Unikraft AMI. We ran the same workload using wrk[1] and then checked the bill at the end of the month, roughly $80 vs $40. [0]: https://aws.amazon.com/marketplace/pp/prodview-xogyq23b3mfge https://aws.amazon.com/marketplace/pp/prodview-xogyq23b3mfge [1]: https://github.com/wg/wrk https://github.com/wg/wrk
qwertyuiop_ 5y ago
There is one primary reason why large companies are migrating to AWS Azure etc that most people seem to be missing here. It’s outsourcing the blame. This is the key to boardroom success. When something goes wrong in their own data centers, the CXOs need to shoulder the blame. Large corporations outsource everything, including the guy who cleans the copiers, copier not working, blame the maintenance vendor. Something went awry with the infrastructure, blame AWS App development project went down the tubes? blame the contractors & outsourcing companies. The Cxo comes out blameless each time
- streamofdigits 5y ago
  that is an argument I have seen before and it is generally true: the core idea is for the corporate entity to operate a riskless arbitrage where all risks are shouldered elsewhere, ideally where they are less understood and thus underpriced. the problem with applying this concept here is the hyperscalers are not exactly a counterparty likely to underprice IT risks. in any case is there any public evidence of this "blame transfer" in action (pressumably with associated remedies)
nickbauman 5y ago
Surprised Cloudflare was mentioned so late in the article and almost a footnote. I think that's not representative of its impact of their strategy on the industry.
- handrous 5y ago
  Judging just from the headline, I assumed the article would mainly be about Cloudflare's entry into that market. The whole point of their strategy has been to position themselves as the foremost middle-men of the Web and then start eating all the service providers "behind" them, including and perhaps especially cloud hosts. They've set everything up and now they're executing on that plan, and things are starting to get interesting.
roody15 5y ago
This is good competition is needed to help with pricing. AWS and other large cloud providers have an incentive to “lock” customers in and raise prices. There must be a pressure by competition to keep this incentive in check.
sirjaz 5y ago
I currently work for a large fortune 50 company as a Senior SRE in charge of server build automation for both cloud and on-prem. We have found that on-prem is cheaper and has less hoops to jump through. We do use vRealize Automation and Orchestration from VMware and it is a good agnostic platform to do this.
- Cthulhu_ 5y ago
  Even when adding the upfront investment costs of hardware and ongoing cost of qualified personnel and support duties? And what if overnight your company needs 10x the capacity?
  - sirjaz 5y ago
    We actually are setup with a major hardware vendor where they roll out that 10x capacity ahead of time, and we only pay for it when we flick on the switch. Otherwise, it is powered off/Standby just sitting there drawing minimal power resources if any.
tzs 5y ago
> Ms Wang and Mr Casado have suggested that firms should think about building their own private clouds to keep costs down. It shouldn't be called a "cloud" in that case. It should be called a "fog".
jillesvangurp 5y ago
I would say this market is due for some consolidation. AWS has basically has enjoyed a nice market share for the last fifteen or so years. In recent years MS and Google have become credible alternatives and in addition to that there are now numerous smaller players that you could feasibly use as well. The key challenges in this space are standardization of commodity features, making the whole business of using any of these solutions less complex, and driving cost down. This goes against how the big three are making of course which is exactly through vendor locking aggravated by complexity. Once you figure out how to use this stuff, you have a lot of non portable stuff depending on that vendor. Replacing that costs money and is never a priority. That's why AWS stuff can be such a PITA to deal with. It actually ensures customers can't move away that easily. However, the common feature set shared by most of these cloud hosting solutions (running vms, databases, docker, load balancers, networking, etc.) is a combination of expensive to use and still too hard to manage. So, the game for new players in the market is being more convenient and less expensive to use. Ultimately private cloud and the cost of getting started as a PAAS provider getting lower means that this is a race to the bottom. Most of the smaller providers already lean heavily on OSS solutions. The question is whether the big three can long term afford to not support the emerging standard stuff. For example at least for me all the convoluted proprietary AWS stuff is actually increasingly a reason for me to avoid depending on them. It was great when they were the only company with decent solutions but that is increasingly less true.
ThinkBeat 5y ago
The choice of running in prem, or even changing cloud vendor is diminishing. Yes if you only use compute and storage moving is easy. If you want scaling in AWS then move to Azure I think that would involve a lot of changes (as it would the other way) I am not sure how much of it Terraform can translate. I would guess K8 can do this is a vendor neutral way but I am not sure since I have never worked with it. I do know for a fact that getting your devops pipeline set up in AWS is really complicated. We ended up having to hire two "elite" AWS consultants for 14 days to get it all working. I was not involved directly in it, but my team had great need of the finished pipeline. If we went with Azure we would have to do it all over again. But most of our stuff is in .Net so getting it all working o Azure would have been magnitudes better. The more "premium"/ high level services one starts using the ability to move on prem or switch cloud becomes harder and harder. Which is of course the goal for the cloud vendors.
- pc86 5y ago
  Speaking from experience, just because Azure is Microsoft doesn't mean that it's point-and-click to get a .NET app running on Azure, especially if you do anything non-standard in startup. Just google "ANCM in-process failure" to see dozens - hundreds - of out-of-the-box .NET apps that just refuse to run in an out-of-box Azure compute/VM.
  - ThinkBeat 5y ago
    I agree. I have been in companies doing both. It is my experience that it is a lot easier on Azure than AWS. (At least as AWS was 2 years ago and Azure one year ago).
- AtNightWeCode 5y ago
  To develop a canonical model over the infrastructure that can be reused between cloud providers takes a lot of knowledge. My experience from Terraform and Azure is that it is not in general worth it.
manishsharan 5y ago
I ,for one, welcome our new IT overlords, the AWS/GCP/Azure. The old overlords of IT, the IT infrastructure depts, were too slow and bureacratic and lacked elasticity. For example, if I wanted to spin up a few servers with Cassandra -- I would have to undergo rounds of review for approval of a new technology then work on requisitioning the h/w or VM, and figure out support chain-of-escalation. Also somehow , on-premise VM were always crashing or running out or disk space. With the new overlords, I have to pass through fewer reviews (I work in financial sector; we love review meetings and gates.) and provisioning a service is as simple as a AWS/GCP/Azure url. And tearing down stuff is easier. The old overlords of IT would be pissed if we decided we did not need a h/w server and we would have to fill out questionaires to determine how we could have been so misguided and then I had to wear my dunce hat for any meeting we would have for the next requisition. I love the battle of cloud computing. With the big three, they are still hold too much power. I look forward to the day when Vultr, DigitalOcean, Ovh, Linode can get approved to be used in my employer's IT stack. With Terraform or Kubernetes, I would be able change our dev/qa/uat cloud environments at will and this will force the Big Three to lower their prices even more.
- 8ytecoder 5y ago
  Not to mention, the upfront budget for hardware or servers were passed to the team requesting a small portion of it if it's the first team to do that.
- jrochkind1 5y ago
  It's interesting that they decided setting up a cloud server did not require the same level of review. That does not seem exactly like the result of a technical difference.
  - manishsharan 5y ago
    When you bring in a new technology or capacity to on-premise, you would want to maximize the bang for the buck which means that you want to review your roadmaps to see who else could benefit from the new technology or capaciy. For Cloud services, this is not really required. Also, there is a level of trust with the Big Three meaning that we can trust that they will be able to manage SQL,Cassandra etc with ease; this IT departmens do not have the manpower bandwidth to keep on adding new servers or services. Hence some of reveiws are not really required.
awinter-py 5y ago
> listing in detail the cost of every single service it has used, from bandwidth in India ($0.01 per request to its website) what does this mean requests != bandwidth. are they paying a penny per request?! that's a lot. or is their website several gigabytes? or are indian CDNs expensive? highly suspicious of economic news service that doesn't understand the numbers it prints

178 points by acqbu 5y ago

CraigJPerry 5y ago

https://archive.md/meT9E https://archive.md/meT9E

asimpletune 5y ago

I wonder when there will be cloud co-ops?

pjkundert 5y ago
Now: https://holo.host https://holo.host Ready for testing and pre-release R&D.
- vgel 5y ago
  IMO, an ideal co-op cloud host wouldn't have a bunch of complicated p2p junk. Just a good ol' datacenter and people to operate it, with the cost and governance shared by the organizations running workloads on it.
- 4ggr0 5y ago
  What's your connection to Holo? You seem to mention them a lot.
  - pjkundert 5y ago
    Yes, I do; sorry about that! They seem relevant, a lot... I was on the development team for a couple years; now, I'm independently developing some technology on the platform. I'll announce it soon, I hope. With the recent AWS outages, and systematic de-platforming of some legal businesses, there seems to be a growing realization that dependence on centralized hosting platforms is ... dangerous. Holo / Holochain seems to provide a unique solution to some of these problems.
    - jatone 5y ago
      pure FUD. bugs happen decentralized systems can be even more problematic to resolve issues in due to their decentralized nature and the usual requirement that fixes be deployed system wide.
    - 4ggr0 5y ago
      Sounds interesting, and thanks for disclosing your connection to Holo! :)
helpfulmandrill 5y ago
Would be much better than getting locked in to a company you have no control over.
praptak 5y ago
The other direction is to use the commercial clouds but aggressively defending their status as a commodity. Have VMs at multiple providers and move your load between them as needed.
streamofdigits 5y ago
This. Turn the "cloud is somebody else's computer" on its head and own part that computer. Why not "own" AWS, GCP, Azure etc.? Because you can't. They are part of conglomerates that come with a moral baggage that is too heavy for some. ps. No "hyperscaling" ability required. The idea that business activity is either complete dominance or death is part of that unfit-for-purpose mentality.
- Layke1123 5y ago
  I do actually own shares in some of those companies. That's kind of the point. You can own the internet, and that's before even "Web 3".
rambambram 5y ago
Cool idea, then we can call it 'the internet' again.
- streamofdigits 5y ago
  its not exactly the same. there are probably significant efficiencies (energy, security etc) in provisioning and sharing "cloud" hardware in reasonably large chunks.
  - rambambram 5y ago
    You are right, but I saw some striking similarities, and needed to joke about it.

k__ 5y ago

Akashnet is an interesting alternative. Sadly, it's Docker based.

imglorp 5y ago
Would you elaborate on the downside? That seems like a common subset that would satisfy a lot of use cases with minimal ceremony.
- k__ 5y ago
  Yes. It's good enough, I guess. I think, a VM approach would have been a bit more flexible.
- itisit 5y ago
  > Akash Network leverages 85% of underutilized cloud capacity in 8.4 million global data centers, enabling anyone to buy and sell cloud computing. [0] How do you even begin to operate a consistent security model across "8.4 million global data centers"? > Auditors on the Akash Network review cloud providers and digitally sign the provider on-chain with their certificate. If you only accept bids from audited providers this means you are trusting the Auditor/Provider not just a Provider. [1] For a business at scale, I don't see this being solid contractual ground. Now I have to trust Auditors to continuously vet Providers? Simply too tenuous for my comfort. [0] https://akash.network/about#vision https://akash.network/about#vision [1] https://docs.akash.network/glossary/security https://docs.akash.network/glossary/security

1cvmask 5y ago

An article that regurgitates Gartner talking points is a tainted one to begin with. Gartner even famously said that Apple should quit the hardware business: https://www.zdnet.com/article/why-does-the-it-industry-continue-to-listen-to-gartner/ https://www.zdnet.com/article/why-does-the-it-industry-conti... https://www.horsesforsources.com/gartner_fail_automation-AI_080418 https://www.horsesforsources.com/gartner_fail_automation-AI_...

klausjensen 5y ago
Twice in my career I have been involved in evaluating enterprise products for a large corporation - and part of that process was looking at Gartner reports. The Gartner reports were complete and utter garbage, filled with factual errors. Looks like they were written by interns, regurgitating marketing material - completely removed from reality. Since then, whenever I see somebody mention a Gartner report, I immediately dismiss that part of the argument.
- bsenftner 5y ago
  Can confirm. Was a developer for 20 years, got an MBA 20 years ago and have been dismayed by the quality of Gartner reports. Worthless for anyone with experience, and dangerous when used to guide the non-technologist. (Yes, I've been developing for 40+ years.)
  - greedo 5y ago
    But mgmt LOVES Gartner reports. It helps them avoid/shift blame when a purchasing decision goes awry. The first question our VPs ask about a new technology is "Is this %VENDOR% in the Magic Quadrant?"
- TheRealDunkirk 5y ago
  I am convinced that Microsoft underwrote them to boost Windows and cast FUD at everything else in the 90's, and they've never recovered their disingenuous ways. What annoys me most are the market share reports, implying that Microsoft owns the entire computing world. They don't. For personal use, I'd say Apple has broken into the majority. I'd pay money out of my own pocket to see PC vs Mac sales numbers with corporate purchases removed, rather than the same report they've published every quarter for 20 years now.

revolvingocelot 5y ago

>Ms Wang and Mr Casado have suggested that firms should think about building their own private clouds to keep costs down Yeah I distantly remember "on prem", too. Does everyone, everywhere really need to be equipped to hyperscale at a moment's notice? Wouldn't it be better not to be beholden to a megacorp that serves you with 30 pages of opaque billing for business-critcal functionality? Doesn't hiring external consultants to manage your fucking cloud infrastructure seem like sort of a risk as compared to keeping the services-wrangler in-house? ...I don't have an MBA, so these are half-serious questions. If anyone can chime in with (at least) a half-serious answer, I'd halfway appreciate it. I understand that The Economist is the voice of global capital, and global capital has decided that there's money in them thar clouds, but still!

arcbyte 5y ago
It's not about hyperscale, it's about time to market. Traditional corporate datacenters were dogshit when it came to provisioning new resources for business capabilities. In today's world, we need the ability to spin up and down deployments very quickly. We don't need huge numbers of resources, just fast response.
- benjaminwootton 5y ago
  I have worked with hundreds of companies moving to AWS, and believe that this is the answer. People are investing in cloud because its faster, more agile and better, not because it's cheaper. The top line looks expensive, but if you then do the correct analysis on the cost side of the equation (taking out staff costs and DC costs) then I think it is either close or in the black from a cost perspective. If the business case was really bad, I don't think we would see the rapid uptake we have seen.
  - hectord 5y ago
    If you think from an economic standpoint, Amazon should price AWS only a bit cheaper than the on-prem options. It's a way to maximize their profits. Then you have competition (GCP, Azure, Alibaba) which will tend to lower the price, but for now the cloud industry grows. Their goal is to attract new customers who didn't use the cloud before.

username223 5y ago

Problem: Managing server hardware takes money and people. Solution: Subscribe to a middle-man that manages it for you. Problem: The middle-man is slowly turning the screws to increase its profits. Solution: Subscribe to another middle-man to figure out how to decrease the first middle-man's profits. It seems to be a universal truth that computing people try to solve all problems with another level of indirection.

yobbo 5y ago
> Problem: Managing server hardware takes money and people. For most, this was never the effective reason. Most rationales behind cloud migrations are likely based on quick scaling, which is meaningless in most cases - except for the signalling of committing to a future where demand hockey-sticks upwards. It's valid if you are netflix, but how many netflixes exist in the world?
- jatone 5y ago
  its not even valid for netflix anymore most likely.
pm90 5y ago
> It seems to be a universal truth that computing people try to solve all problems with another level of indirection. This isn’t about computing but about specialization. Should McDonalds be in the business of operating Data Centers? Or should they just buy it from those who do it best and focus on selling food?
- baq 5y ago
  i don't know. should a book store be?
  - pm90 5y ago
    It should not.
- username223 5y ago

jrochkind1 5y ago

> Mr Garman counters that the higher price of moving data off a cloud (“egress” in the jargon) reflects the higher costs of that exercise. Almost by definition, customers leave with more data than they entered with. Uh, they charge more per byte, that's not what's going on. I don't know if that was AWS's Garman being deliberately misleading, or the Economist journalist misunderstanding.

jasode 5y ago

>Businesses’ main motive for moving to the cloud was never about cost but “scalability”: having access to additional computing resources with a few clicks. That's not accurate because many businesses do migrate for cost reasons. Especially for internal non-public facing servers like email. E.g. They moved from self-hosted Microsoft Exchange servers to cloud-managed email like Office 365 Exchange Online -- or switch to another cloud competitor like G Suite. >Ms Wang and Mr Casado have suggested that firms should think about building their own private clouds to keep costs down. So far few firms have opted for such “repatriation”, which is both costly and makes it harder for businesses to enjoy the benefits of essentially unlimited computing resources in the public cloud. As a case study, The Guardian newspaper tried building their own private cloud and failed. They ultimately just went with AWS: https://web.archive.org/web/20160319022029/https://www.computerworlduk.com/cloud-computing/guardian-goes-all-in-on-aws-public-cloud-after-openstack-disaster-3629790/ https://web.archive.org/web/20160319022029/https://www.compu... If the company's business profit is not in the tech sector (like The Guardian), this means its IT datacenter operations is not its core competency which makes it hard to build internal IT that can keep up with AWS capabilities. Tech businesses like Dropbox, Facebook, can hire & maintain skilled IT staff like AWS but most other non-tech businesses can't.

rmbyrro 5y ago
They say the "main motive", not the sole motive. Cost is also a motive, for sure, but do you consider they'd move to reduce costs if there wasn't scalability?
krab 5y ago
> The Guardian newspaper tried building their own private cloud and failed What I see often is that it's hard and costly to replicate the whole cloud experience in-house. OpenStack is a behemoth to operate. However, if you build something more specific to your needs (and often mix in selected public cloud services), you'll get a much cheaper stack. Small businesses can often run very efficiently with a few rented physical servers in a colo, G Suite and AWS S3. Larger ones can provide a platform specific to the technologies used in-house and combine with public cloud for any other tech stacks.

eole666 5y ago

> The situation is analogous to a century ago, when electricity became an essential input You can do nearly everything "the computing cloud" offers without it : just use servers, install your services on it, manage it, invest in more servers/ more powerfull servers if needed... Cloud computing is not a brand new invention, it's just delegating the server and services management to others... Whereas, a century ago, electricity was actually an essential input.

londons_explore 5y ago
In the future, I can imagine cloud providers offering discounted bandwidth from one AWS customer to another. Then I can imagine SAAS companies offering discounts if your API requests originate from AWS cloud. Maybe even Amazon prime members will get free bandwidth from their home internet connections into and out of AWS. At some point, the whole lot becomes a walled garden, and being outside it is more expensive if you need to use services that are inside.
- rmbyrro 5y ago
  Cost of SaaS or API prices are usually orders of magnitude higher than their infrastructure marginal costs. Although this idea makes practical sense, from a financial standpoint I don't think there will be enough incentive. Bandwidth savings in infra will represent <1% savings relative to the SaaS/API price.
martincmartin 5y ago
Companies used to run their own electrical generators. That seems analogous to running your own servers.
- mschuster91 5y ago
  > Companies used to run their own electrical generators. ... and have the staff with experience and resources to keep them running. These days, UPS maintenance is sourced out to the cheapest bidder and testing is infrequent, with the predictable result of shit hitting the fan at the exactly worst time.

robertlagrant 5y ago

Either The Economist or I don't know what Snowflake is.

DangitBobby 5y ago
I had that same thought at first, but I think they were trying to do a tie in for enterprise SaaS. They mention Hashicorp as well.

exdsq 5y ago

My first job was IT support for on-prem systems and, as I battle my containerized deployment scripts this morning, really miss those days. Probably just nostalgia but I had named them all after superheroes and it was just way more fun when you got to spec a server and install it, fix them (physically) when they had issues, and panic when the AC broke. Edit: Also I was way more fit from climbing around a building all day cabling stuff. Genuine question - does anyone know what sort of IT jobs require physical work? I miss those days a lot!

VTimofeenko 5y ago
> does anyone know what sort of IT jobs require physical work? I have seen a few positions in the US for datacenter technicians. Those positions had explicit requirements(probably because of ADA) to be able to lift over X lbs as part of day to day activities.
marvinblum 5y ago
Hetzner would probably be the perfect working space for you. They build all of their servers by hand on custom racks. You can move around in 14+ buildings in Falkenstein, manage cables, fix servers, work on their cloud offering, ...
- teloli 5y ago
  AND you would get to live in Falkenstein, one of the most exciting places in Europe.
  - mellavora 5y ago
    which does indeed sound like a good home for servers named after superheros.
imadethis 5y ago
COVID has probably changed this, but networking jobs tend to require a lot of racking and cabling work, especially if you’re fitting out a new work area. Even with the majority of your infrastructure moved to the cloud, you’ll still need APs or Ethernet jacks wired into a LAN for the end users.

jamesliudotcc 5y ago

Also, security. Has anyone noticed that all of the big ransomware cases have been for on-prem servers, not cloud?

betaby 5y ago
Not really. On other side news like 'my AWS account racked 40000$ while I was asleep (because of the crypto-miners)' make to the front page way to often.

nderjung 5y ago

One way to reduce compute costs significantly is by using a much lighter stack. Typical deployments to the cloud are based on the traditional OS model, and more often than not are a container-in-a-VM solution to deploy some service. The reality here is that there is so much unused, unnecessary and wasted compute resources. Why deploy a full OS and kernel which will only ever serve static web content, for example. And then there's the attack surface that comes with a full OS, running additional services like SSH which may only ever be used once. Major vendors like Amazon and Google are happy to provide this to anyone and with simple on-boarding, but, as is clear in the article, it's expensive. With Unikraft[0], you can completely re-package your application as a unikernel, automatically. It's a single VM image and does nothing more than just running the application. It boots straight to the main application, it doesn't have a shell, and has nothing else running along side it. The result is a huge cost saving (up to 50%), since you're no longer having to pay for these extra resources just to run the a full traditional OS image plus your application. [0]: https://unikraft.io https://unikraft.io

mikkelam 5y ago
Maybe you should also mention that you are affiliated with this project. This reads like an ad.
- nderjung 5y ago
  Whoops :) Yes I am affiliated, I can't update my comment but I'll make sure to indicate this if/when I write a comment which discusses anything about Unikraft.
pid-1 5y ago
50% compared to what? AWS EC2? Lambda? Google Cloud Run? Those sort of numbers without even having a pricing page give me 0 confidence sorry.
- nderjung 5y ago
  We are launching a pricing page in the next month as we are gearing up for launch :) One example of recieving a 50% cost saving is from Amazon Machine Images (AMIs). We compared the cheapest official NGINX AMI available on the Amazon Marketplace[0] and then also ran an NGINX+Unikraft AMI. We ran the same workload using wrk[1] and then checked the bill at the end of the month, roughly $80 vs $40. [0]: https://aws.amazon.com/marketplace/pp/prodview-xogyq23b3mfge https://aws.amazon.com/marketplace/pp/prodview-xogyq23b3mfge [1]: https://github.com/wg/wrk https://github.com/wg/wrk

qwertyuiop_ 5y ago

There is one primary reason why large companies are migrating to AWS Azure etc that most people seem to be missing here. It’s outsourcing the blame. This is the key to boardroom success. When something goes wrong in their own data centers, the CXOs need to shoulder the blame. Large corporations outsource everything, including the guy who cleans the copiers, copier not working, blame the maintenance vendor. Something went awry with the infrastructure, blame AWS App development project went down the tubes? blame the contractors & outsourcing companies. The Cxo comes out blameless each time

streamofdigits 5y ago
that is an argument I have seen before and it is generally true: the core idea is for the corporate entity to operate a riskless arbitrage where all risks are shouldered elsewhere, ideally where they are less understood and thus underpriced. the problem with applying this concept here is the hyperscalers are not exactly a counterparty likely to underprice IT risks. in any case is there any public evidence of this "blame transfer" in action (pressumably with associated remedies)

nickbauman 5y ago

Surprised Cloudflare was mentioned so late in the article and almost a footnote. I think that's not representative of its impact of their strategy on the industry.

handrous 5y ago
Judging just from the headline, I assumed the article would mainly be about Cloudflare's entry into that market. The whole point of their strategy has been to position themselves as the foremost middle-men of the Web and then start eating all the service providers "behind" them, including and perhaps especially cloud hosts. They've set everything up and now they're executing on that plan, and things are starting to get interesting.

roody15 5y ago

This is good competition is needed to help with pricing. AWS and other large cloud providers have an incentive to “lock” customers in and raise prices. There must be a pressure by competition to keep this incentive in check.

sirjaz 5y ago

I currently work for a large fortune 50 company as a Senior SRE in charge of server build automation for both cloud and on-prem. We have found that on-prem is cheaper and has less hoops to jump through. We do use vRealize Automation and Orchestration from VMware and it is a good agnostic platform to do this.

Cthulhu_ 5y ago
Even when adding the upfront investment costs of hardware and ongoing cost of qualified personnel and support duties? And what if overnight your company needs 10x the capacity?
- sirjaz 5y ago
  We actually are setup with a major hardware vendor where they roll out that 10x capacity ahead of time, and we only pay for it when we flick on the switch. Otherwise, it is powered off/Standby just sitting there drawing minimal power resources if any.

tzs 5y ago

> Ms Wang and Mr Casado have suggested that firms should think about building their own private clouds to keep costs down. It shouldn't be called a "cloud" in that case. It should be called a "fog".

jillesvangurp 5y ago

I would say this market is due for some consolidation. AWS has basically has enjoyed a nice market share for the last fifteen or so years. In recent years MS and Google have become credible alternatives and in addition to that there are now numerous smaller players that you could feasibly use as well. The key challenges in this space are standardization of commodity features, making the whole business of using any of these solutions less complex, and driving cost down. This goes against how the big three are making of course which is exactly through vendor locking aggravated by complexity. Once you figure out how to use this stuff, you have a lot of non portable stuff depending on that vendor. Replacing that costs money and is never a priority. That's why AWS stuff can be such a PITA to deal with. It actually ensures customers can't move away that easily. However, the common feature set shared by most of these cloud hosting solutions (running vms, databases, docker, load balancers, networking, etc.) is a combination of expensive to use and still too hard to manage. So, the game for new players in the market is being more convenient and less expensive to use. Ultimately private cloud and the cost of getting started as a PAAS provider getting lower means that this is a race to the bottom. Most of the smaller providers already lean heavily on OSS solutions. The question is whether the big three can long term afford to not support the emerging standard stuff. For example at least for me all the convoluted proprietary AWS stuff is actually increasingly a reason for me to avoid depending on them. It was great when they were the only company with decent solutions but that is increasingly less true.

ThinkBeat 5y ago

The choice of running in prem, or even changing cloud vendor is diminishing. Yes if you only use compute and storage moving is easy. If you want scaling in AWS then move to Azure I think that would involve a lot of changes (as it would the other way) I am not sure how much of it Terraform can translate. I would guess K8 can do this is a vendor neutral way but I am not sure since I have never worked with it. I do know for a fact that getting your devops pipeline set up in AWS is really complicated. We ended up having to hire two "elite" AWS consultants for 14 days to get it all working. I was not involved directly in it, but my team had great need of the finished pipeline. If we went with Azure we would have to do it all over again. But most of our stuff is in .Net so getting it all working o Azure would have been magnitudes better. The more "premium"/ high level services one starts using the ability to move on prem or switch cloud becomes harder and harder. Which is of course the goal for the cloud vendors.

pc86 5y ago
Speaking from experience, just because Azure is Microsoft doesn't mean that it's point-and-click to get a .NET app running on Azure, especially if you do anything non-standard in startup. Just google "ANCM in-process failure" to see dozens - hundreds - of out-of-the-box .NET apps that just refuse to run in an out-of-box Azure compute/VM.
- ThinkBeat 5y ago
  I agree. I have been in companies doing both. It is my experience that it is a lot easier on Azure than AWS. (At least as AWS was 2 years ago and Azure one year ago).
AtNightWeCode 5y ago
To develop a canonical model over the infrastructure that can be reused between cloud providers takes a lot of knowledge. My experience from Terraform and Azure is that it is not in general worth it.

manishsharan 5y ago

I ,for one, welcome our new IT overlords, the AWS/GCP/Azure. The old overlords of IT, the IT infrastructure depts, were too slow and bureacratic and lacked elasticity. For example, if I wanted to spin up a few servers with Cassandra -- I would have to undergo rounds of review for approval of a new technology then work on requisitioning the h/w or VM, and figure out support chain-of-escalation. Also somehow , on-premise VM were always crashing or running out or disk space. With the new overlords, I have to pass through fewer reviews (I work in financial sector; we love review meetings and gates.) and provisioning a service is as simple as a AWS/GCP/Azure url. And tearing down stuff is easier. The old overlords of IT would be pissed if we decided we did not need a h/w server and we would have to fill out questionaires to determine how we could have been so misguided and then I had to wear my dunce hat for any meeting we would have for the next requisition. I love the battle of cloud computing. With the big three, they are still hold too much power. I look forward to the day when Vultr, DigitalOcean, Ovh, Linode can get approved to be used in my employer's IT stack. With Terraform or Kubernetes, I would be able change our dev/qa/uat cloud environments at will and this will force the Big Three to lower their prices even more.

8ytecoder 5y ago
Not to mention, the upfront budget for hardware or servers were passed to the team requesting a small portion of it if it's the first team to do that.
jrochkind1 5y ago
It's interesting that they decided setting up a cloud server did not require the same level of review. That does not seem exactly like the result of a technical difference.
- manishsharan 5y ago
  When you bring in a new technology or capacity to on-premise, you would want to maximize the bang for the buck which means that you want to review your roadmaps to see who else could benefit from the new technology or capaciy. For Cloud services, this is not really required. Also, there is a level of trust with the Big Three meaning that we can trust that they will be able to manage SQL,Cassandra etc with ease; this IT departmens do not have the manpower bandwidth to keep on adding new servers or services. Hence some of reveiws are not really required.

awinter-py 5y ago

> listing in detail the cost of every single service it has used, from bandwidth in India ($0.01 per request to its website) what does this mean requests != bandwidth. are they paying a penny per request?! that's a lot. or is their website several gigabytes? or are indian CDNs expensive? highly suspicious of economic news service that doesn't understand the numbers it prints