45

I understand the concept of cloud computing, but I'm curious why the term has become so exhausted the past several years. Servers have been around for a long time, and I fail to see how this is any different from before the term "cloud computing" was in fashion. There are many more vps services and more systems and complexity, but

is "cloud computing" mainly just a marketing term?

6
  • 3
    Your question is very interesting and can be answered objectively. But the way you express it makes it sound very opinion-based. I steongly suggest to reformulate and explain what you mean with “marketing term” and to what you oppose it, in order to facilitate objective answers :-)
    – Christophe
    Sep 9 '20 at 6:36
  • 7
    Refer to the NIST definition (link) of five essential characteristics of cloud computing: (1) on-demand self-service, (2) broad network access, (3) resource pooling, (4) rapid elasticity or expansion, and (5) measured service. Refer to the full content for definitions. Overall claims of cost reduction, reliability, or general superiority are often not objectively substantiated. However, there are stories of highly resilient and fault-tolerant implementations that mention episodes that
    – rwong
    Sep 9 '20 at 7:11
  • 2
    ... that mention episodes that are typically not survivable (e.g. large scale regional disasters and destruction of backup and cold storage facilities) if they were implemented using the organization's own resources without using today's cloud computing services.
    – rwong
    Sep 9 '20 at 7:13
  • 4
    Is "ride-sharing" mainly just a marketing term? After all, you've been able to buy cars for over a hundred years.
    – Tom W
    Sep 9 '20 at 18:31
  • 2
    Cloud computing is the contemporary term for time sharing. And inexplicably opaque billing systems.
    – copper.hat
    Sep 10 '20 at 4:51
114

The distinguishing feature of "cloud computing" is indeed the way that it is marketed, in particular, the way that it is priced.

Another synonym for "cloud computing" that I personally prefer is "utility computing", and that term describes best what it is all about: it is priced and used like any other utility, water, gas, electricity.

You only pay for what you use, when you use it, you don't have to configure anything, you don't have to rent anything, you don't have to prepay anything. You are automatically billed monthly based on your very fine-grained actual usage.

It really is like a utility: if you want to wash your hands, you open the tap a little bit and a bit of water comes out. If you are filling your pool, you open the tap more and more water comes out. You don't have to prepay the water, you don't have to call the water company and ask them to send you water, you don't have to arrange anything. You just open the tap, and there is instant water.

Utility computing resources are the same way.

This is different from anything we had before. We had rented servers in data centers, but we had to pay those whether we used them or not. Even in the (very short and unsuccessful) era of Application Service Providers (ASPs, anybody remember those), you generally had a monthly or yearly plan. There were mainframe sharing systems where you were billed by the CPU second, but those weren't as instantaneous as utility computing resources, you generally had to pre-arrange some stuff.

And in the field of economics, the sub-field that deals with how to assign prices to products, and how to bring those products to the market is called "marketing", so you are almost right: "cloud computing" is mainly a marketing term, but I would very much object to the word "just" in your sentence:

is "cloud computing" mainly just a marketing term?

Because the marketing aspect of cloud computing is precisely what makes it different from everything that came before, and what made it so disruptive.

There are other parts of the "utility" metaphor that are also applicable to utility computing, such as the fact that you don't need to care where your water comes from and how it gets to your tap, you just turn on the tap, and water comes out. The water could come from a tank, a reservoir, a lake, a river, a well. The electricity could be generated by wind, solar, geothermal, coal, nuclear, it could come directly from a plant owned by your provider or by a plant owned by a different provider who then sold the energy to your provider, etc.

This is where the "cloud" term comes in. It comes from system diagrams, where the network was always drawn simply as a "magical cloud" that does everything, and you don't really need to concern yourself with how it works. That is the metaphor that "cloud computing" is meant to invoke. The cloud is just this thing that is always there, always works, and you don't need to worry about it.

13
  • 3
    I happen to have studied Information Economics before I dropped out of university, right around the time of the decline of ASPs and rise of cloud, and our professors had us study this deeply, because it was a prime example of what this new field of study that nobody could precisely explain what it was for, was about: the intersection of economics and informatics. Sep 9 '20 at 10:59
  • 3
    « You only pay for what you use » is very relative: in a SaaS you may only pay for the users you have registered. but not every user uses all the features you are paying for. In an IaaS you may pay for servers of a predefined capacity, regardless of the really used compiting power, i.e you pay for idle time as well. And some suppliers even sell you a SaaS offer in a private cloud for which a sizing must be done and which binds the price to that sizing (which is in reality a traditional hosting scheme).
    – Christophe
    Sep 9 '20 at 14:39
  • 17
    @Christophe: I would argue that at some point, it then simply stops being a cloud and is a classical hosting service, no matter what the vendor's advertising brochure may say on its title. Elasticity and utility-like pricing are part of the defining feature set of what it means to be "cloud", so if you don't have that, then it's not a cloud. It's perfectly fine not being a cloud, of course, nothing against that. A "cloud" offering that isn't elastic is about as "cloud" as the Democratic People's Republic of Korea is a democracy and cares about its people. Sep 9 '20 at 14:46
  • 1
    "Don't need to care where it comes from and how it gets there." No, you quite often do need to care about this when dealing with any serious usage of cloud computing, especially if you want to provide services truly worldwide (and yes, I'm saying this based on personal experience). Sep 9 '20 at 22:17
  • 5
    "This is different from anything we had before." - Those with enough gray hair will remember time-sharing systems, which share many of what you consider key characteristics of cloud computing. You logged in from a remote terminal and paid for the time you were connected and how much CPU you consumed. Of course, there were no VMs in those days and the no choices of CPU speed, ram, disk. You just got some share of the machine. MULTICS (MIT) and DTSS (Dartmouth) were among the earliest examples. TSS later ran on IBM 360 hardware and TOPS-20 on DEC-20 machines.
    – Llaves
    Sep 10 '20 at 20:38
9

The term “cloud” is indeed very broad and used to represent different realities:

  • IaaS is “Infrastructure as a service”. This seems to be your current way to see the cloud: its about computing capacity and servers.
  • PaaS” which is “platform as a service”
  • SaaS is “Software as a service” which gives you access to a software without any care about the servers or the platform.

You are right when you question the if cloud is a technical reality that would make it different from other technologies: technically, you can run your own server, install some platforms and operate your own software and provide it to your users directly. Very often, this can be done with the same technology and stacks.

You can even add internal firewalls, load-balancing, high-availability, geographical distribution in your own premises. If you have enough money (e.g. if you’re the pentagon), you can build up your own private cloud, and no expert, that would analyse your architecture, could make the difference.

So cloud is indeed a commercial reality: it’s about who is doing what and at what price, which is defined in a contract. So there is a difference between owning and operating the things yourself, and let others do it for you, for a price, but with a know how and a flexibility that you might never acquire if selling cloud is not your main business.

That’s the way amazon started this business (and maybe IBM before, with its former on-demand offers).

In conclusion, cloud is more defined by the contractual terms than the technology. Now where it comes tricky is the IaaS, where there is an overlap with commercial hosting offers. Here I have seen relatively inflexible hosting offers sold as “cloud” but without the expected flexibility in practice. So here “cloud” is sometimes used as a “marketing term” which correspond neither to a commercial nor a technical reality.

Finally, be aware that the commercial reality has impact on your architectural possibilities. So even if there are no differences in the technology, the distribution of roles will deeply impact what you can or cannot do on your side.

4

Cloud Computing in a marketing term from an angle and not just a marketing term from another angle.

End User Perspective: For the end users it doesn't matter if you run your work loads in cloud or on your premise or in any other way. From the end user point of view it really doesn't matter.

So if this term is used with the end user, it would be close to a marketing term.

Development organization Perspective: When your workloads are running in a cloud instead of on premise, there are several benefits

  1. The cost of maintaining the servers on your premise will not be incurred.
  2. The organization doesn't need to have people skilled to manage servers
  3. With cloud its easy to have elastic load. Many businesses have varying loads, with cloud its easy to manage it. It would be time consuming to provision and de-provision servers

Consideration to go on cloud:

  1. Now you should make sure your code can run with making assumptions in cloud. So while developing the code has to be development in a cloud native way (without making assumptions about deployment environment)
  2. It would be good if the code is written to take advantage of the elasticity the cloud offers. This is a skill highly in demand
  3. You need to have the skill set in the organization to deploy the code in the cloud. It's not trivial. You might have 100s of work loads and you might need some automation in terms of CI/CD.
  4. Cloud is an umbrella term. When you are going to use it you need to know the details such as service offered by your Cloud Vendor and competitors. This is also a skill set.

So from the perspective of an organization moving to cloud, it's not just a marketing term. The organization needs to be ready with the necessary skill sets

2
  • There's one more critical consideration for me personally. The cost of upgrading servers (hardware) in cloud computing is zero or even negative (AWS often release newer, faster hardware at a lower price than previous generation). Without this ability I don't consider it a cloud service but merely a hosting service
    – slebetman
    Sep 10 '20 at 1:18
  • yea - i agree this answer is missing Business owners perspective, the CAPEX, OPEX is a huge difference to why organizations are going for "cloud computing" over on prem servers... Sep 10 '20 at 20:26
0

Servers have been around for a long time

Yes, servers. But cloud computing it's not about servers, it's a business model that didn't exist before, this is why deserves it's own terminology and it's not just a marketing term but the biggest technological revolution of the last 25yrs, at least

Yes, today everybody talks about AI, ML and how much these things are cool. Will computer make us slave whithin next 5 years?

Think about, though, that AI is a very old concept. We already knew from decades that it was possibile, we knew that a computer could predict stocks value and tell cats from dogs in photos, but how to do that in a PC with a pentium IV CPU? The answer could be use remote supercomputers for calculations and deploy models on web services, that is less or more what we do today.

Let's say that you have a business idea: you create a mobile app where users can insert their photos and see how they look like when they become old. You know that you can do that with neural networks. In your business plan you state that you will need to build a supercomputer and the service will cost 100$ per photo. Turns out that your business is not sustainable

Today, thanks (?) to cloud computing, this kind of application could be sold for free (at leat apparently, anyway with a very litlle unitary cost)

After all, the cloud is the evolution of VPS concept. In the past, servers where physical assets that users can manage. Then virtualization, Moore's law, devops, High bandwith internet connections/private fiber backbones along with a bunch of other tech progresses made the jump possible. I think that very few non-IT people have got the impact of this change in our society and global economy

8
  • 5
    How does cloud computing stop your service from costing $100 per photo?
    – user253751
    Sep 9 '20 at 15:48
  • 2
    Because you don't build a supercomputer, you just rent time from virtualized resources that run on crappy hardware shared among thousands of other companies. You don't need an ops team to keep that infrastructure healthy but just set up automated tools that interact with cloud provider, they will take care of many things.you can even save to develop AI algorithm, there are managed services for that. You will just build applications that leverage cloud benefit and run your business,with very low ingress barriers Sep 9 '20 at 22:12
  • You mean, the cloud lets you rent 10% a supercomputer for 10% of the time if that's what you need? (except it doesn't, really. You often have to buy one-hour blocks)
    – user253751
    Sep 10 '20 at 10:42
  • 1
    Cloud computing is not necessarily cheaper that owning your own infrastructure. Once you're in the petabytes-of-storage category, it's really not cheaper. There's a lot of rah-rah cheerleading here that glosses over that it's only applicable to a specific kind of use case. Sep 11 '20 at 16:29
  • 1
    ...I will say that I've worked at two places where we saved large amounts of money (even after accounting for the costs of geographic replication) by rolling our own physical infrastructure, and when I joined both, they were still startups. Sep 11 '20 at 17:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.