Want to point you to the just out article in The Economist "Tanks in the cloud".
Ron Wolf
Content includes market sizings for the 3 CC tiers.... SaaS - $12B/yr;
PaaS - $300M/yr; and IaaS - $1B/yr. Of these they find the Iaas market
to be the most difficult to estimate and the most interesting part of
the article is how they go about doing this. This is partly based on a
(perhaps misleading, or flawed or both) derivation by Guy Rosen of
Cloudkick, based on instance 'serial numbers', that 90,000 new virtual
instances a day are started on AWS. Yikes, that seems high to me. And
as one commenter on The Economist website points out, this doesn't
take into account how many instances are killed every day (indeed, the
method could not derive this). In any case, the # of new instances per
day shows a pleasing exponential growth....
http://www.economist.com/node/17797794
Apologies if the article isn't public, as a subscriber, its hard for
me to know.... In any case, in a comment to the article Mr. Rosen
generously posted a link to his dataset:
http://www.jackofallclouds.com/2010/12/recounting-ec2/
Perhaps you have comments on the market sizings or methods used to
derive them? For instance, I find it fascinating that the PaaS tier,
as pushed by Google & MS, is lagging dramatically. Will this continue?
Or is PaaS just off to a slower start? Maybe a case of a great
architecture (IMO) but a lack of customers with matching needs?
Also I have a vital question, does anyone really pronounce SaaS as
'sarse'? Maybe I'm just out of it, but I would feel pretty dumb saying
'sarse' , 'parse' (in this context), or, 'eye-arse' (OMG!). But before
I post back to The Economist accusing them of being half-arsed (ha hoo
hoo), thought to check with y'all.
Amy Wohl
In any case, I think that their statistical method seems flawed -- too much
speculating on unknowables and, as pointed out, nothing in their for
cancelled sessions in an environment where huge numbers of test sessions
exist for only minutes or hours.
The comments are almost as interesting as the article.
Amy
Amy D. Wohl
Editor, Amy Wohl's Opinions
1954 Birchwood Park Drive North
Cherry Hill, NJ 08003
856-874-4034
a...@wohl.com
www.wohl.com
http://www.economist.com/node/17797794
http://www.jackofallclouds.com/2010/12/recounting-ec2/
--
~~~~~
UP 2010 Conference: http://www.up-con.com
UP-Start Cloud Awards http://www.up-con.com/awards
Official Cloud Slam websites - http://cloudslam.org , http://cloudslam10.com
and http://cloudslam09.com
Posting guidelines: http://bit.ly/bL3u3v
Follow us on Twitter http://twitter.com/cloudcomp_group or @cloudcomp_group
Post Job/Resume at http://cloudjobs.net
Buy hundreds of conference sessions and panels on cloud computing on DVD at
http://www.amazon.com/gp/product/B002H07SEC,
http://www.amazon.com/gp/product/B002H0IW1U or get instant access to
downloadable versions at http://cloudslam09.com/content/registration-5.html
and http://cloudslam10.com/content/registration
~~~~~
You received this message because you are subscribed to the Google Groups
"Cloud Computing" group.
To post to this group, send email to cloud-c...@googlegroups.com
To unsubscribe from this group, send email to
cloud-computi...@googlegroups.com
Andrew Badera
Bernard Golden
discussions regarding open source software. There were people (and in
fact, there still are people) who, for example, would dismiss the role
of MySQL because it had a far lower revenue stream than Oracle. This
completely misses the point.
When a vastly cheaper alternative enters a market, the total revenue
of the new entrant is much smaller than the incumbent. Two phenomena
occur: (1) price elasticity decisions by users increase total use of
the new, less expensive alternative; and (2) people find new uses for
the less expensive alternative. Point in case, does anyone believe
that data-driven web sites would number in the millions if each of
them had to have an Oracle database driving them? And, would the new
social media/groupon, etc. (indeed, even Google itself) have come into
being if high functionality, data-driven, low-cost websites weren't
assumed to be part of the technical landscape? The net effect,
ultimately, is that the new alternative revenues eventually outstrips
the incumbent enormously. Earlier in the cycle, however, it just looks
like the new alternative is puny because its revenues are much smaller
than the existing product.
A more relevant comparison is total proportional use within a market,
i.e., what percentage of all database installations does MySQL
represent? Extended to the cloud computing market, what proportion of
all computing is performed on public IaaS infrastructure, and, even
more crucially, what is happening to the trend of proportional use?
Miha Ahronovitz
Economist is not very "deep" in trying to estimate these revenues, in fact Randy Bias - whom we both know - supplied most information to Economist staff members.
The point you make about MySQL (whom I know relatively well from my when Sun still existed), that the technology quietly expands via being affordable, and market share revenues come later - is true, but many times is too late. The same happened to OpenSolaris, and look at MySQL, owned by Oracle . Simply Oracle has no interest to make the cheaper alternative widely available, as it will erode it's own customers base.
IMO in IaaS is not much room for open source, but there is a lot of room for "pure IaaS", it means public infrastructures not loaded with proprietary APIs as Amazon, Google, etc are loaded.
The $1B estimated in Economist for IaaS is peanuts versus the total real market. I like your suggestion. If I start a new IaaS market for a start up, the investors want to see 20 large customers who will buy your service. (or products). There is no way - as far as I know - to identify actual names of all prospective customers. This is why analysts will always estimate markets, their estimates will be different from one another. Their work is very helpful,because is all we have for now.
So we must count clouds, as if they are trains as in the World War 2 the way the statistical espionage operated..
Kindly see (sorry repeating) my blog, which I researched this very subject, including Economist
http://my-inner-voice.blogspot.com/
--- On Sun, 1/9/11, Bernard Golden <bernard...@gmail.com> wrote:
Amy Wohl
new entrant is much smaller than the incumbent.
Yes. But what counts is whether having a vastly cheaper alternative stirs
up new uses or new users, in which case the total market size will become
much (perhaps exponentially) larger and much of that growth may be based on
the cheaper alternative. Why don't you start with PCs vs. mainframes and go
from there? On the day the IBM PC was announced I had a one-on-one call
with IBM about their new product (I couldn't get to the press announcement)
and they assured me the total market for PCs was 5,000. It might be more
relevant to compare PCs to minicomputers since mainframes are still around
but minis are largely gone as a result of the rise of the PC market.
Amy
Amy D. Wohl
Editor, Amy Wohl's Opinions
1954 Birchwood Park Drive North
Cherry Hill, NJ 08003
856-874-4034
a...@wohl.com
www.wohl.com
-----Original Message-----
From: cloud-c...@googlegroups.com
[mailto:cloud-c...@googlegroups.com] On Behalf Of Bernard Golden
Sent: Sunday, January 09, 2011 6:36 PM
To: Cloud Computing
Subject: [ Cloud Computing ] Re: Want to point you to the just out article
in The Economist "Tanks in the cloud".
--
Gilad Parann-Nissany
Regards
Gilad
__________________
Gilad Parann-Nissany
Ron Wolf
plainly in front of our nose still needs clear restatement - at least
for me.
i take two points of view.
investor PoV - investor's can think long and/or short. on the long
side, Elison's comment (think this was in comments to The Economist
article) to the effect of "Clouds are only water vapor" is both
brilliant and stupid at the same time. brilliant because he knows his
customer base and knows that they crave this kind of pandering. I know
that it may raise hackles to say this, but Oracle DB was never the
best technical or ROI choice. Oracle Corp is is now far more than DB,
but most of it is still based on the pre-cloud licensing model. on the
short and. regarding Elison, stupid side, i'm reminded of Drucker's
observation to the effect of "businesses biggest mistake is to over-
invest in the bread winners and under-invest in new products and
markets." will Oracle be the next GM?
but, as small fry investor and a techie at heart, i much more think of
my own career, what is best for my clients, and have always been drawn
to new and, i hope, effective technologies. and in that regard,
@Bernard you comments are quite encouraging and, i think, spot on,
thx,
___________Ron
Rich Wellner
It might be more relevant to compare PCs to minicomputers since mainframes are still around but minis are largely gone as a result of the rise of the PC market.
I don't disagree with this, but since we're talking economics... I found it very interesting that one of my MSP customers, back when I was with Univa, bragged about AS/400 being, by far, their most profitable hosting offering.
rw2
Amy Wohl
Yes. One of the realities of Cloud Computing is that using zillions of PC servers isn’t necessarily the best hosting solution for everyone. Some highly automated minis like the AS/40 can do a good job and IBM is putting in mainframes in some cases.
Amy D. Wohl
Editor, Amy Wohl's Opinions
1954 Birchwood Park Drive North
Cherry Hill, NJ 08003
--
Bernard Golden
I believe Oracle and GM comparison is apt in one dimension, but not
entirely applicable in another. GM was out-innovated by other car
companies that took share, but it was a market that was (relatively)
static in terms of unit size, so it's nearly a zero-sum game for
manufacturers. Also, switching costs for GM users were relatively low
-- wait a few years and buy another mark; in other words, low lock-in.
Oracle represents enormous lock-in, as software decisions are
typically multi-decade.
However, the unit size of IT-based and IT-enabled products is
increasing, perhaps at an exponential rate, as a phenomenon never
before present in economies (Moore's Law) continues to apply within
processors.
In sum, I don't think Oracle will experience a GM near-death
experience. The stickiness of its lock-in (and the unbelievable
profitability of packaged software) precludes that (btw, the book
Overhaul by Steven Rattner that discusses the GM and Chrysler
bankruptcies from the inside of the Obama administration is absolutely
riveting). However, it will come to occupy a progressively smaller and
smaller tip of the market pyramid as less expensive technologies like
IaaS come to the fore. Of course, one could get a far more graceful
and compelling description of this dynamic in "The Innovator's
Dilemma." :-)
Frank Greco
XaaS terms. If you think of a many-tiered Cloud stack, it makes so much
more sense then 3 simpleton layers (eg, is AWS+Beanstalk IaaS or PaaS?,
is Eucalyptus PaaS or an abstract IaaS? Where does RightScale fit in
the XaaS strata? And what the heck is Azure trying to be? etc...)
The market size projection from the Economist (like Nostradamus, they're
not always right) in interesting. It makes sense that SaaS is the
biggest market, since the customer base are "real" end-users, not
"unreal" developers (which of course most of us here are). Its the
applications stupid(tm). PaaS is smaller because its for us geeks. And
its smaller because all the PaaS vendors want to lock you into their
frameworks. See... IT has not changed in 30 years. Its the same
lock-em-up strategy that IBM employed back in the non-3D-flat-panel-TV
days of yore, when Frank thought "mythical man month" meant a
"mythical-man month" celebration and threw a party for Sasquatch, the
Yeti, the Abominable Snowman and John Wayne during the start of my
freshman year. The PaaS vendors are all trying to position themselves
as THE solution to your development woes. So there's a bit of market
chaos which causes confusion, so there's less of a desire for customers
to dive in. Its the classic "What if we pick the wrong API Harry?"
"Well then we're screwed dude" "Then you pick first" "Nope... you
pick" "Nope... you" deadlock...
As to pronunciation of "aaS", I would suggest one thing: don't.
Frank G.
- The definition of Cloud Computing is "IT 2.0"
hmmm... no... strike that...
The definition of Cloud Computing is "IT 2.2.1_35b"
On 1/9/11 2:59 AM, Ron Wolf wrote:
> Intriguing title, eh? I think worth the time (a few minutes) to read.
> Content includes market sizings for the 3 CC tiers.... SaaS - $12B/yr;
> PaaS - $300M/yr; and IaaS - $1B/yr. Of these they find the Iaas market
> to be the most difficult to estimate and the most interesting part of
> the article is how they go about doing this. This is partly based on a
> (perhaps misleading, or flawed or both) derivation by Guy Rosen of
> Cloudkick, based on instance 'serial numbers', that 90,000 new virtual
> instances a day are started on AWS. Yikes, that seems high to me. And
> as one commenter on The Economist website points out, this doesn't
> take into account how many instances are killed every day (indeed, the
> method could not derive this). In any case, the # of new instances per
> day shows a pleasing exponential growth....
>
> http://www.economist.com/node/17797794
>
> Apologies if the article isn't public, as a subscriber, its hard for
> me to know.... In any case, in a comment to the article Mr. Rosen
> generously posted a link to his dataset:
>
> http://www.jackofallclouds.com/2010/12/recounting-ec2/
>
> Perhaps you have comments on the market sizings or methods used to
> derive them? For instance, I find it fascinating that the PaaS tier,
> as pushed by Google& MS, is lagging dramatically. Will this continue?
Greg Pfister
Miha Ahronovitz
I commented the Economist article on my blog
http://bit.ly/ePCpBf
To answer your questions
On 2/18/2011 8:17 PM, Frank Greco wrote:
> First... thank you to the marketing bozos who invented the meaningless
> XaaS terms.
The marketing bozos invented everything in the computer world, so the
people can understand & buy in.
> If you think of a many-tiered Cloud stack, it makes so much more sense
> then 3 simpleton layers (eg, is AWS+Beanstalk IaaS or PaaS?,
It is PaaS, AWS specific PaaS
> is Eucalyptus PaaS or an abstract IaaS?
It is PaaS for private clouds
> Where does RightScale fit in the XaaS strata?
Special purpose PaaS
> And what the heck is Azure trying to be? etc...)
Azure is PaaS no one likes to use.
Pure IaaS: I create a server, I install my OS as user and I install my
application, just as I do o na physical server I just bought
Pure SaaS: I use the application directly as a user.
Anything in between pure IaaS and pure SaaS that others impose on me as
developer is PaaS. The PaaS ideally should be based on free will, not
forced-feed in our throats as if we were geese.
Miha
Frank Greco
On 2/20/11 3:51 AM, Miha Ahronovitz wrote:
> On 2/18/2011 8:17 PM, Frank Greco wrote:
>> First... thank you to the marketing bozos who invented the
>> meaningless XaaS terms.
> The marketing bozos invented everything in the computer world, so the
> people can understand & buy in.
Yep. I agree. Marketing drives the industry, not us.
But not all marketing is quality stuff. Just like good
architecture and crap architecture, great restaurants and
Zagat rejects, NOAA and the Weather Channel, there's a similar
good<->bad spectrum in marketing.
>> If you think of a many-tiered Cloud stack, it makes so much more
>> sense then 3 simpleton layers (eg, is AWS+Beanstalk IaaS or PaaS?,
> It is PaaS, AWS specific PaaS
>> is Eucalyptus PaaS or an abstract IaaS?
> It is PaaS for private clouds
>> Where does RightScale fit in the XaaS strata?
> Special purpose PaaS
I think you're agreeing with me Miha. :)
Once you start to differentiate the types of each aaS, then you've
broken the marketing model, right?
Imho, we need a stack before any semblance of true *utility*
computing could ever possibly exist, if that
truly is a goal for us... maybe it isn't, but if it is, the aaS
model ain't going to do the trick.
> And what the heck is Azure trying to be? etc...)
> Azure is PaaS no one likes to use.
Run Windows on EC2 and you have Azure.
> Pure IaaS: I create a server, I install my OS as user and I install
> my application, just as I do o na physical server I just bought
> Pure SaaS: I use the application directly as a user.
>
> Anything in between pure IaaS and pure SaaS that others impose on me
> as developer is PaaS. The PaaS ideally should be based on free will,
> not forced-feed in our throats as if we were geese.
The qualifier "Pure" is clearly subjective. Are you in marketing
Miha? ;)
Frank G
- clouds/mobile is next up at the plate, but who is the on-deck
hitter?
miha ahronovitz
The qualifier "Pure" is clearly subjective. Are you in marketing Miha? ;)
By "pure" IaaS I mean I load an OS, then I load the application, exactly the same as in a physical computer. Assume that someone says': STOP!!!, you need and AMI. What is an AMI? I am told , Oh, it is something very simple. Simple? Perhaps. But I need to learn something I never used before. It is something that AWS forces upon me, without asking my opinion. In this moment, I have no longer pure IaaS. I have AWS. And the AMI is useless if I want my application to run on a non-AWS cloud
Note that they are more and more pure IaaS providers. CloudSigma is one of them.
Miha
Frank Greco
On 2/26/2011 12:28 PM, Frank Greco wrote:The qualifier "Pure" is clearly subjective. Are you in marketing Miha? ;)I have news: we are all in marketing, the moment we post on this group
there are more levels in the cloud strata than the original simple 3-layer cake.
By "pure" IaaS I mean I load an OS, then I load the application, exactly the same as in a physical computer. Assume that someone says': STOP!!!, you need and AMI. What is an AMI? I am told , Oh, it is something very simple. Simple? Perhaps. But I need to learn something I never used before. It is something that AWS forces upon me, without asking my opinion. In this moment, I have no longer pure IaaS. I have AWS. And the AMI is useless if I want my application to run on a non-AWS cloud
If someone asks you "Is Amazon an IaaS vendor?", how do you answer?
With all due respect, I think you are agreeing with me that the simple [IPS]aaS model is not cutting the mustard.
Note that they are more and more pure IaaS providers. CloudSigma is one of them.
interface so true cross-vendor *utility* cloud computing can eventually be a reality. Or perhaps utility computing
is further out or just a dream due to vendor competition. Or maybe we're forcing the implementation of the
electricity analogy because we don't know better yet (ie, see "birds v planes").
Just an aside... in your opinion, what is the different from a "pure" (your word) IaaS cloud provider and
a hosted service that has fine-grained metering? Just curious.
Frank G.
- "SaaS??? All of this cloud stuff is software. The applications are software, the frameworks are software and
even the hardware is software."
Miha Ahronovitz
There is more than one truth in the world, co-existing. Your point is that "aaS is a silly-stack". So maybe you are right. It may be silly, but it is the best we have so far. Here is a twit that summarizes my opinion.
"What is PaaS? Anything between pure IaaS & pure SaaS that others force-feed in our throats as if we were geese, not cloud consumers"
Your argument is all cloud is software. Yes, but the access to this software is different: Users see only SaaS. Developers must know the PaaS to be able to host services. And the IaaS providers must know how to offer instances that behave most closely to real hardware. The audience for each layer is different.
Other question of your is "Is Amazon an IaaS vendor?" If you read the above, it is simple. You can not buy pure IaaS from AWS. You have to buy the whole stack , with their Hadoop, their RD, their AMI. It is like wanting to buy a truck list priced at 20K, but it is not available. The only truck you can buy is the 60K all bells and whistles, selected by others as best for you.
To answer your question: "Is Amazon an IaaS vendor?" The answer is NO. The AWS is an Iass + PaaS package vendor, take it or leave it.
Other question from you:
"what is the different from a "pure" (your word) IaaS cloud provider and a hosted service that has fine-grained metering? Just curious."
Look at hosting provider RagingWire http://ragingwire.com/ . Click right from the home page to Stratscale, their cloud service subsidiary. Try to configure their server VM offers
https://store.stratascale.com/ss/index.php/public-cloud/virtualscale-level-1-vm-server-bundle.html
See there the minimum term. Is 1 month. Sure they can make it 5 minutes, but then they will blow off the entire hosting portfolio that gives them immense revenues compared to a cloud billing. It is like shooting yourself in the foot.
So from now on, let's continue on private. You may have a point in that aaS-silly-think, but I still don't know why. I respect you right to label silly anything you want. I won't buy it, others may. It would be sad all of us to agree, so disagreeing is a form of progress.
miha
From: Frank Greco <fgr...@javasig.com>
To: miha ahronovitz <mij...@sbcglobal.net>; cloud-c...@googlegroups.com
Sent: Sun, February 27, 2011 5:52:43 PM
Subject: Re: [ Cloud Computing ] Want to point you to the just out article in The Economist "Tanks in the cloud".
Follow us on Twitter @cloudcomp_group
http://www.amazon.com/gp/product/B002H07SEC, http://www.amazon.com/gp/product/B004L1755W, http://www.amazon.com/gp/product/B002H0IW1U or GET instant access to downloadable versions at
- http://www.up-con.com/register
- http://cloudslam09.com/content/registration-5.html
- http://cloudslam10.com/content/registration
Khazret Sapenov
Miha Ahronovitz
http://www.cloudsigma.com/en/platform-details/the-api
You have APIs to create, start, pause or kill a server or a group of servers. IMHO one can use these API to create a logical decision to shut-down servers using these APIs
Miha
Khazret Sapenov
Miha Ahronovitz
I asked your question to Robert Jenkins, CloudSigma's CTO, and here is his reply:
You can easily scale using our API and a monitoring service like Appfirst and we have a mobile application developer doing this already. It isn't complicated for anyone looking to do this as, if they are running such infrastructure they are more than capable of throwing together the script necessary. Saying that it is something that someone can publish and make available to others in the future or stick a web front end on too. Regardless it already is possible within your own controlled infrastructure with relative easy (a significant percentage of our customer do it and do it in a highly tailored way as per our usual position)
So this expansion shrinking that we call elasticity is, like eau-de-cologne or wine, a personal taste, depending on customer's needs. Rather than ruthlessly imposing one solution, teh cloud provider must not sub estimate, but respect the intelligence of the customers.
Miha
Khazret Sapenov
Miha Ahronovitz
This is what I call complete freedom for the customer to select the PaaS they need for their cloud. You could see the cloud offerings jcloud, or enStratus runing on cloudsigma.
The idea that AppFirst an go our of business is possible (after all Lehman Bros went out of business or merged into oblivion, DEC went out of Business, Compaq and Lotus123,) However the argument is not relevant here. As long as they exist, they are a valid option, together with many more.
Nothing is for ever. Not even the Universe :-)
Miha
Khazret Sapenov
Khazret, if you have such an autoscale solution, contact CloudSigma and they can offer it as an option to customers, the same way they offer AppFirst now.
This is what I call complete freedom for the customer to select the PaaS they need for their cloud. You could see the cloud offerings jcloud, or enStratus runing on cloudsigma.
The idea that AppFirst an go our of business is possible (after all Lehman Bros went out of business or merged into oblivion, DEC went out of Business, Compaq and Lotus123,) However the argument is not relevant here. As long as they exist, they are a valid option, together with many more.
Nothing is for ever. Not even the Universe :-)