Re: Cloud Computing with Octave

[Georgios Kousiouris]

Jordi Gutiérrez Hermoso <address@hidden> said:

> On 19 October 2012 15:52, Georgios Kousiouris <address@hidden> wrote:
> > And it is very convenient to reply in a dogmatic way to the 1% of the
> > response, without actually answering the other parts.
> 
> You begin with flawed premises, so it's just a matter of attacking the
> flawed premise, namely that "cloud computing" is a meaningful term.
> Once the flawed premise is dealt with, the rest of your argument which
> hinges on that premise does not follow, so it's not necessary to
> attack it carefully. From a false assumption anything follows.

In order to prove the flaw of the premise, you must provide arguments and not
simply state that cloud computing is not a valid term.

> 
> But fine, as you wish. Let's do this, let's assume that cloud
> computing is meaningful. Let's begin with a few attempted defintions
> on the internet:
> 
> Wikipedia:
> 
>     Cloud computing is the use of computing resources (hardware and
>     software) that are delivered as a service over a network
>     (typically the Internet).
> 
>     http://en.wikipedia.org/wiki/Cloud_computing
> 
> From SearchCloudComputing:
> 
>     Cloud computing is a general term for anything that involves
>     delivering hosted services over the Internet. These services are
>     broadly divided into three categories: Infrastructure-as-a-Service
>     (IaaS), Platform-as-a-Service (PaaS) and Software-as-a-Service
>     (SaaS).
> 
>     http://searchcloudcomputing.techtarget.com/definition/cloud-computing
> 
> From Infoworld:
> 
>     Cloud computing is all the rage. "It's become the phrase du jour,"
>     says Gartner senior analyst Ben Pring, echoing many of his peers.
>     The problem is that (as with Web 2.0) everyone seems to have a
>     different definition.
> 
>    
http://www.infoworld.com/d/cloud-computing/what-cloud-computing-really-means-031
> 
> From GNU:
> 
>     The real meaning of “cloud computing” is to suggest a
>     devil-may-care approach towards your computing. It says, “Don't
>     ask questions, just trust every business without hesitation. Don't
>     worry about who controls your computing or who holds your data.
>     Don't check for a hook hidden inside our service before you
>     swallow it.” In other words, “Think like a sucker.” I prefer to
>     avoid the term.
> 
>     https://www.gnu.org/philosophy/who-does-that-server-really-serve.html
> 

That is why standardization organizations such as NIST exist
(http://csrc.nist.gov/publications/nistpubs/800-145/SP800-145.pdf) and issue
definitions, so that this disambiguation is resolved.

> Let us assume that despite the difficulties that all of these have in
> defining cloud computing, we can infer a definite meaning at any rate.
> If that is so, the only meaning I'm able to infer is that there is a
> network connection and the remote end is performing some computing
> tasks. But this is the case of absolutely any computing network, they
> all have to communicate with each other and the different ends do
> specific computing tasks. So this cannot be the intended meaning that
> we can deduce from all four of these definition.s Let us assume, then,
> that everyone else on the internet is wrong but you're not.

This simplification may be further lowered to "everything is electric signals
".So computers are finally the same as kitchen appliances or any electrical
device.

> 
> Let us be charitable to you and instead use your own definition:
> 
> > "Cloud computing" means combination of virtualization issues,
> > resource sharing and scheduling issues, concurrency and
> > multi-tenancy issues, performance interference at cpu,network and
> > storage, and that's just at the infrastructure layer. At the
> > software and platform layers there are many more even more
> > challenging aspects like elasticity, application scaling, data
> > integrity etc.
> 
> You have conflated quite a few differing topics into cloud computing
> that the other definitions haven't, but let us assume that unless
> there is virtualisation, resource sharing, scheduling (more generally,
> concurrency), then it's not cloud computing.

This is not a definition, this is an argument to indicate the difficulty of
cloud computing implementation and management that may in turn prove that it
is indeed a technology, given that it does not simply apply concepts from
other fields, but combines, introduces new challenges and thus innovation. The
combination of these factors is what makes it challenging and not what is
defining it. 

> 
> So, given your definition, which of the following are cloud computing
> and which are not? After all, the point of a definition is to set
> limits, to "de-fine" what you want to talk about. My attempt to answer
> this question follows.
> 
>    * Remote virtual machines hosted as a service -- Yes, a virtual
>      machine has an OS, and any modern OS has concurrency, resource
>      management, and a measurable performance, so it's cloud
>      computing.
> 
>    * Google docs -- Probably, Google probably uses a virtual machine
>      and uses an OS.
> 
>    * A web server serving only static pages -- Only if the server
>      itself is hosted as a virtual machine. If there is no
>      virtualisation, then it's not cloud computing.
> 
>    * An IRC daemon: Yes, but only if the daemon is hosted in a virtual
>      machine, but there must be other daemons also running in this
>      virtual machine so that the other elements of cloud computing
>      (resource sharing and application scaling) can also exist.
> 
> I am not sure your definition actually works here, but I think it does
> at least determine the cloudness of these four examples.

As stated above, the topics mentioned were not part of a definition but of
research challenges for building a well performing cloud. If you want to
actually know if these examples can be considered cloud computing, you should
try to match them to the NIST definition.

> 
> Let us treat your other arguments independently of your definition of
> cloud computing, in order to avoid the possible danger of arguing from
> a false premise.
> 
> > Have you ever been given an account on a server or system in general
> > without root privileges? I guess you have, I have plenty of times,
> > either as a student or afterwards. Did you have ability to control
> > that system? Did you use that system even though it was out of your
> > control? Should you have the ability to control it? What if you
> > didn't have the necessary knownledge and did something that messed
> > up the entire configuration?
> 
> I have been a guest in other people's machines, yes. This is just a
> network service. My hosts have always used free software, and given me
> enough resources to install and run my own software so I have never
> felt restricted while I was their host. I independently have trusted
> my own hosts because they are usually my friends.

To me this is a trust issue. you can always have friends with clouds
> 
> > What is more, there is also the case of a private cloud
> 
> This term needs to be defined. Does that simply mean a private network
> server or does it also need all of the elements of cloud computing you
> included above?

If you want to use the basic IaaS setup, this means that you will have a
number of physical hosts running specialized middleware (there are a variety
of open source alternatives, OpenStack, OpenNebula, Eycalyptus etc.) and offer
the virtual resources (to your trusted friends) through an API.
> 
> > run by a single entity (e.g. company or university) that may give
> > the capability to its members (employees or students) to use these
> > software and hardware resources as a service, without having to know
> > (in most cases they don't want to know either) anything about what
> > is going on in the backstage. Isn't this helpful?
> 
> Encapsulation is helpful only if it can be sidestepped by those who
> don't need it.

There are people that actually do not need or want to know this information.

> 
> > Is there a trust issue involved?
> 
> Depends on the individuals and entities in this relationship.

Exactly. So if you have for example university students accessing the cloud
for a lab exercise there is no trust issue for their code.
> 
> > Isn't the code you write as an employee of a company already on an
> > SVN or something similar?
> 
> I assume you're trying to argue about the trust that exists between
> an employee and an employer. Yes, there is usually some trust there,
> but not a whole lot. Because people don't trust each other, they draft
> contracts to be enforced through legal means. This is disappointing in
> some cases. I do not see what this has to do with the meaning and
> utility of "cloud computing", unless again you are conflating all
> network use with clouds.

This refers to the issue of making the code available to the company you are
working for (which refers in turn to your argument regarding running something
somewhere that you do not have control). It is logical to assume that if you
work for a software company you would actually deliver the code you are
working on for a company product. So running it on the company cloud is not a
trust issue. The company will anyhow obtain the code in the end. 
> 
> > Furthermore, should all of the users be computer experts in order to
> > have access and usage of scalable resources or exotically installed
> > applications?
> 
> If they want to be treated like suckers, they can act like suckers and
> accept sucker terminology like "cloud computing".

So an  e.g. doctor would also need a computer science degree to perform
medical simulations. 
> 
> > So even if you see it from a business model point of view, cloud
> > computing gives the democratic ability
> 
> There is usually no democracy here; we are not all publicly voting
> what to do with each other's servers.

Democracy has the point that anyone is able to advance from an equal starting
point. If you see
> 
> > to the individual to actually matter and create something from
> > scratch without having the extended knowledge or capital that would
> > otherwise be needed to compete with existing, long standing
> > companies.
> 
> So you see "cloud computing", whatever that is, as enabling
> disadvantaged people from participating in computing. The extent to
> which this is true is doubtful, 

Why? see example with doctors. Another example: if you are a standalone
researcher, how will you find the resources to run a 50-node map reduce job
for example? How much time and cost would it take you to actually build it?
Are you interested in building it, given that your main interest is around
e.g. data analytics? 

>but it seems to me like you are
> promoting a serfdom of technically incapable people because they are
> unable to do their own computing on their own machines.

how about capital expenses? will i need many servers? cooling systems? energy
systems? how about time-to-market?  
Technical incapability may mean that someone is efficient in one area and not
in all areas. Using an API is much simpler than having to do all the work from
scratch.

Anyhow, let's not annoy people any more, if you want we can continue this
discussion in private emails.

BR,
george

> 
> > Again, I am not arguing on using Cloud computing for everything.
> > There are cases where it offers significant benefits and other cases
> > where it has drawbacks (security-wise or otherwise). It is all up to
> > the use case, but discarding the technology as a whole is not the
> > answer.
> 
> I insist that "the technology" is a vague and meaningless term and
> that cloud computing by itself does not constitute a technology, but
> according to your own definition, constitutes a variety of different
> technological advances that in isolation are interesting (concurrent
> programming is interesting, network topology is interesting,
> virtualisation is interesting), but to bundle all of these together
> under your definition as a "technology" is a broad and overly general
> term, which leads to overly broad statements that confuse the
> individual cases.
> 
> - Jordi G. H.
> 
> 



--