I think it should read 'cloud computing makes some servers obsolete'.
Not everybody will be able to host 'in the cloud', the kind of data they host could easily be forbidden to pass to third parties, even for hosting. Think of medical data, banking and so on.
Quite a few servers could probably be hosted 'in the cloud', but for now the cost benefits are not really there unless your application falls in to a very specific niche.
Bandwidth and storage premiums in the cloud make it very tough to position a cloud based solution vs hosting your own stuff. The only case when it makes sense is if you need large numbers of servers for a short period and if you are growing faster than you can order hardware.
It's also great as a fall-back plan and to create redundancy.
You have a good point, but he at least partially addressed this in the article when he said "First, I really like the use of the article “a” in reference to cloud as it speaks to all models of cloud: private, public, external, internal, and hybrid."
A company that cannot or will not put their data on the internet and "The Cloud" could at least in principle build "A Cloud" within their own intranet and host there, still gaining the flexibility of focusing on applications rather than servers.
Still, I think you have a good point with cloud computing makes some servers obsolete. Some data not only cannot leave the company intranet, but must be segregated strictly from other data within in the same company. This will require focusing on the server.
Also, it can (in some cases) be harder and slower to write in such a flexible than to let it be tied to a single server. Doing that makes no sense if programmer time is a precious asset in that organization and you will have a need for massive scaling in the foreseeable future.
There's also a lot of software which (for now) licenses by the core or physical socket and will not run (technically and/or legally) in a cloud or virtualized environment.
Sounds nice in theory, in reality less than 1% of applications really need to scale and only one "resource" a.k.a. "server" is going to have to do all the work, simply for economic reasons. Current (shared) PaaS offerings still have too many limitations and lock-in to be a suitable replacement.
Not everywhere. Many rural markets still use them even in America, to say nothing of other countries.
New technology almost always forces the old into a niche, not completely removing it. There is still a genuine demand for buggy whips for instance, its just a small and specialized one now.
Yes you are. You don't have to think about the servers, the cloud abstracts it away. Here's an analogy. High level languages make memory management obsolete. You could then ask, "because programs are made up of something other than memory and instructions?", and you'd definitely be missing something.
Your analogy makes sense, but it is as broken as the central idea in this piece, and they're both so for the same reason:
Despite high-level languages, memory management is still very much a reality for many programmers. People still write ASM and C every day.
Memory management and physical boxes may be obsolete for Joe Schmoe, but he's not the guy who makes it possible to ignore these things in the first place. There are all sorts of folks out there who's choice to gain copious amounts of low-level knowledge has allowed me to be lazy.
I disagree with your implications that someone is lazy if they have concerns other than low level bit twiddling. Some of us are trying to build businesses or simply progress the state of the art in other areas. For example, I work in a research group on information retrieval. I could spend hours of my day making progress on IR with higher level languages or I could instead waste productivity on manual memory management. Choosing the former is not being lazy.
Also, the idea that "some" people still do C therefore memory management isn't obsolete for most people is clearly flawed.
I meant the good kind of lazy, so yeah, I wasn't really trying to insult you.
Also, the idea that "some" people still do C therefore memory management isn't obsolete for most people is clearly flawed.
As flawed as the absolute claim, "Cloud Computing Makes Servers Obsolete". Not only is it objectively false (servers are necessary for cloud computing, duh), but it makes no attempt to quantify by saying for whom they've become obsolete.
I admit to not reading the article (based on the ridiculous title), but if it claims that cloud computing makes skills in setting up dedicated hardware from OS up and then engaging in smart capacity planning and proper deployment schemes so as to make horizontal (or vertical) scaling as painless as possible without the advantages provided by virtualization less necessary in today's computing landscape, then...
"loud computing makes skills in setting up dedicated hardware from OS up and then engaging in smart capacity planning and proper deployment schemes so as to make horizontal (or vertical) scaling as painless as possible without the advantages provided by virtualization less necessary in today's computing landscape"
that isn't a good title. Given the intended audience, you can assume the readership is smart enough to understand that nuances are not reflected in the title.
Here are some examples of using your style in well known titles.
God is not Great Most of the time (worse title, but more accurate)
A failure of Capitalism to self regulate banking and insurance and avoid systemic risk (much worse than the original, but more accurate)
A pale blue dot when viewed from a distant perspective in space (much worse than the original, but more accurate)
I guess my basic point is that people talk about the "cloud" like it is a magical entity that solves all our problems when it comes to server load, resources and data. The reality is, it's still hardware, it's still physical, and it is still a limited resource.
Not having to think about something doesn't change the nature of the thing you aren't thinking about. Clouds are still built from servers.
It does change the nature of the thing to the person using the cloud. I agree with your points about the silver bullet hype, but the cloud does give people a lot of advantages and a lot of cost savings under the right circumstances. Those circumstances happen to include a very large segment of web developers.
Twenty years ago I was reading about how networked workstations made mainframes obsolete, this is just more of the same. Network latency effects, if nothing else, is going to ensure the continued need for at least some centralized computing resources.
All of the fawning over the cloud really irritates me. I use S3 and EC2 and love them both. However, I also still use local storage and non-EC2 servers.
Everything has its place, and the cloud has definitely not replaced the server for me, nor do I see that happening any time in the near future. Especially if you're just starting out, dealing with getting set up in a cloud environment is a headache you just don't need.
Cloud computing provides service abstraction at different levels.
You don't need to think about servers at SaaS (Software as a Service. e.g, Google Docs/Apps) and PaaS (Platform as a Service. e.g. Google Apple Engine, Windows Azure; Amazon S3, Elastic MapReduce etc.) level, but you do have (virtualized) servers when you need IaaS (Infrastructure as a Service. e.g. Amazon EC2, VMware infrastructure) where you need to put existing/legacy applications that have clear dependencies on explicit servers in the cloud.
28 comments
[ 2.7 ms ] story [ 80.7 ms ] threadNot everybody will be able to host 'in the cloud', the kind of data they host could easily be forbidden to pass to third parties, even for hosting. Think of medical data, banking and so on.
Quite a few servers could probably be hosted 'in the cloud', but for now the cost benefits are not really there unless your application falls in to a very specific niche.
Bandwidth and storage premiums in the cloud make it very tough to position a cloud based solution vs hosting your own stuff. The only case when it makes sense is if you need large numbers of servers for a short period and if you are growing faster than you can order hardware.
It's also great as a fall-back plan and to create redundancy.
A company that cannot or will not put their data on the internet and "The Cloud" could at least in principle build "A Cloud" within their own intranet and host there, still gaining the flexibility of focusing on applications rather than servers.
Still, I think you have a good point with cloud computing makes some servers obsolete. Some data not only cannot leave the company intranet, but must be segregated strictly from other data within in the same company. This will require focusing on the server.
Also, it can (in some cases) be harder and slower to write in such a flexible than to let it be tied to a single server. Doing that makes no sense if programmer time is a precious asset in that organization and you will have a need for massive scaling in the foreseeable future.
New technology almost always forces the old into a niche, not completely removing it. There is still a genuine demand for buggy whips for instance, its just a small and specialized one now.
Am I missing something here, or is this just semantics?
"let’s start right now and focus on the application as the central entity to our architectures, not the physical server"
Despite high-level languages, memory management is still very much a reality for many programmers. People still write ASM and C every day.
Memory management and physical boxes may be obsolete for Joe Schmoe, but he's not the guy who makes it possible to ignore these things in the first place. There are all sorts of folks out there who's choice to gain copious amounts of low-level knowledge has allowed me to be lazy.
Also, the idea that "some" people still do C therefore memory management isn't obsolete for most people is clearly flawed.
Also, the idea that "some" people still do C therefore memory management isn't obsolete for most people is clearly flawed.
As flawed as the absolute claim, "Cloud Computing Makes Servers Obsolete". Not only is it objectively false (servers are necessary for cloud computing, duh), but it makes no attempt to quantify by saying for whom they've become obsolete.
I admit to not reading the article (based on the ridiculous title), but if it claims that cloud computing makes skills in setting up dedicated hardware from OS up and then engaging in smart capacity planning and proper deployment schemes so as to make horizontal (or vertical) scaling as painless as possible without the advantages provided by virtualization less necessary in today's computing landscape, then...
I agree wholeheartedly!
That makes for a great title doesn't it? LOL
How about, Cloud Computing Making Dedicated Hardware Obsolete For Some. Was that so difficult? Took me 10 seconds.
Here are some examples of using your style in well known titles.
God is not Great Most of the time (worse title, but more accurate)
A failure of Capitalism to self regulate banking and insurance and avoid systemic risk (much worse than the original, but more accurate)
A pale blue dot when viewed from a distant perspective in space (much worse than the original, but more accurate)
Not having to think about something doesn't change the nature of the thing you aren't thinking about. Clouds are still built from servers.
Everything has its place, and the cloud has definitely not replaced the server for me, nor do I see that happening any time in the near future. Especially if you're just starting out, dealing with getting set up in a cloud environment is a headache you just don't need.
So when your boss is screaming at you "why are we down?!", can you get by with just saying "because X (the cloud) is down". Highly unlikely.
You don't need to think about servers at SaaS (Software as a Service. e.g, Google Docs/Apps) and PaaS (Platform as a Service. e.g. Google Apple Engine, Windows Azure; Amazon S3, Elastic MapReduce etc.) level, but you do have (virtualized) servers when you need IaaS (Infrastructure as a Service. e.g. Amazon EC2, VMware infrastructure) where you need to put existing/legacy applications that have clear dependencies on explicit servers in the cloud.