Post to Twitter |
Comments (20) |
Views (53075) |
There is an interesting debate going on over on the Google cloud computing group that also helps point out some of the appropriate use cases for cloud computing. The example used is a simple comparison of Amazon EC2 vs. purchasing a set of servers for development purposes ( I have added some additional costs and scenarios below ) This example also assumes the servers fit in existing space and either environment would be managed by existing staff.
| |
Purchase - on Premise |
|---|---|
| $ 15,000 |
Quad-Core Servers ( 5 x 3,000 each ) |
| $ 750 |
1/2 Rack + Gigabit Switch |
| $ 15,750 |
Total Hardware cost |
| $ 5,800 |
Annual amortized cost, 5% over 3 years |
| $ 0 |
Assuming no incremental real estate cost |
| $ 2,000 |
Annual power & AC cost |
| $ 7,800 |
Total annual cost on premise |
| Purchase - at Colo |
|
|---|---|
| $ 8,000 |
Colo fee's; 1/2 Rack + power + bandwidth |
| $ 5,800 |
Annual amortized cost |
| $ 13,800 |
Total annual cost at Colo |
| Cloud |
|
|---|---|
| $ 35,040 |
24x365x5 Amazon EC2 ( $.80 per high CPU Server instance hour ) |
| $ 8,320 |
40 hours x 52 weeks |
| $ 688 |
40 hours x 4.3 weeks |
On the surface it's apparent that EC2 is significantly more expensive if the set up is utilized 24x7x365, even a 40 hour week yields a slightly higher cost. So where is all the savings ? What's all the hype about ? This simple example does point out that the Cloud is not always a more cost effective solution it really comes down to what is the particular use case and alternative costs. For example if there is no space available or the existing space has reached the power limits of the facility ( a more common occurrence ). That means that the likely scenario is finding a Colo facility to provide space power and bandwidth. Depending on location and bandwidth usage this could easily cost $8,000+ per year plus additional remote administration hardware and service fees, effectively increasing the annual cost of purchased equipment to near $ 14,000. Although this option is still less than Amazon if utilized 24x7x365, it now is significantly more than the cost of the 40 hour week at EC2 which may be reality for a development environment. And if you only need the setup for a month of dev or testing Amazon becomes a no brainier.. put on your credit card !
What both examples point out are the fact that there is no single answer. In fact the right answer for many companies might be premise plus cloud. In order for this to work for a single workload however a seamless connection would be required, recognizing this has led to the Citrix Cloud Bridge based on our WANScaler acceleration technology. In fact, Citrix is in the unique position to be able to assemble the prerequisite technologies that make the C3 Citrix Cloud Center an optimized solution for many scenarios.
There are many other pro's, con's and hidden costs of each option, I am interested to hear what the community has considered regarding Cloud economics and/or other factors.
Comments (20)
Sep 27, 2008
Michael Keen says:
Chris, PERFECT!!! I've been watching the threads over on Goog...Chris,
PERFECT!!! I've been watching the threads over on Google Cloud Computing Group too and was thinking about this. I have been saying it over and over to folks that cloud doesn't fit every scenario and it's not for everyone. Cloud computing is the way that things are going but that isn't for another five years or so. I know of some companies that are staking their future on the fact that everyone will be moving to the cloud sooner rather than later. Now there are some great use cases for using the cloud; bursting to the cloud to cover an increase in resources, test/dev, DR, etc. But to move mission critical workloads, I don't think we are even close to being there yet. Not just from a cost perspective either, there are other issues that need to be hammered out like security, management, etc.
Let me cover the one scenario I'm working on right now; Disaster Recovery to the cloud. I think this is the first step of getting people to move into this new computing model. The sheer technical complexity and overhead has always made DR out of reach for most customers, until now. I can't share the revenue model with you here unfortunately but it's compelling. The big value proposition to the customer is a complete DR solution but the CapEx cost for hardware and software is eliminated and they now have a much smaller monthly OpEx. C3 is playing a critical part in the architecture. It was playing a critical piece even before Citrix released Cloud Center.
Thanks for posting this up Chris. I'll keep you up-to-date on my progress and I'll be happy to share the architecture once it's completed.
Cheers
Michael Keen
Sep 28, 2008
Chris Fleck says:
Michael, thanks for the insightful comment. I agree DR is a great fit for a Clou...Michael, thanks for the insightful comment. I agree DR is a great fit for a Cloud solution. Existing Disaster Recovery solutions are prohibitively expensive due to the duplication of hardware and infrastructure. The potential for the Cloud to reduce these intrinsic costs is huge. Please do keep us informed of your plans.
Sep 27, 2008
Anonymous says:
Good analysis , we have been doing the same calculations. When you add the time ...Good analysis , we have been doing the same calculations. When you add the time factor of justifying, buying and setting up servers to the equation it makes the cloud look even better. Right now that means dev and test for us.
Anonymous replies:
Sep 28, 2008
Anonymous says:
The "Cloud" is the latest marketing re-packaging for an old concept. Whether you...The "Cloud" is the latest marketing re-packaging for an old concept. Whether you call it the "Cloud" or "Grid" or ASP, it is all the same thing.
You left out the biggest cost that corporations want to minimize - labor. That is what the "Cloud" is all about - slashing labor costs. But it really isn't a surprsie you failed to mention it. No one wants to talk about what the "Cloud" really means - reducing expenditures on payroll and benefits, pensions and health care for IT personnel. If you cannot out source the labor to India or some other distant land, then outsource the entire operation to the "Cloud". It sounds so perfectly harmless when you call it the "Cloud". A "Cloud" couldn't hurt anyone, could it?
Technology companies like Citrix, VMWare, Microsoft, Dell, IBM and HP all hope to convince corparations to cut expenditures on labor and spend those dollars on the "Cloud" so those tech companies can get that money in software, hardware and service sales. More than a few CIO's will be willing to move a bit of thier operations in this "Cloud" in the sky in hopes of cutting labor costs and improving the bottom line. And more than a few readers of this blog and other tech blogs will lose their jobs to the "Cloud".
As more and more CIO's put their head and budget into the "Cloud", NASDAQ will creep up. Options and bonuses will rain down from the "Cloud" for a while.Tech execs will be ditching comercial first class for private jets again and be getting double the expense account (including the one for their spouse).
But this "Cloud" does not have a silver lining. The "Cloud" will be filled with over hyped products built and tested by mostly out sourced developers who could care less about the final product. Patches will be released with ridiculous time outs (sound familiar?) or obvious bugs that will poison the entire "Cloud". Security will be about as difficult to penetrate as a - cloud. Personal data will slowly drizzle from the "Cloud". One day, there will be a torrential downpour and the "Cloud" will pour down like acid rain and wash away into the sewer.
No worries though. Soon tech companies will come up with a new metaphor, and CIO's will have a new euphemism for cutting the jobs of the very people who built their entire infrastructure and keep it running day to day. And the whole cycle will begin again.
I wonder what you will call it next time around.
Sep 28, 2008
Chris Fleck says:
Thanks for your perspective. Your comment brings up some additional challenges t...Thanks for your perspective. Your comment brings up some additional challenges that "Cloud" adoption will face both in terms of the impact on IT issues such as security and resources but also the impact on IT personnel. It is true that the "Cloud" does offer the opportunity to save on labor costs, however the same could be said of most innovations of the industrial age. In fact many IT jobs where probably justified and created by replacing manual or paper driven work done by others. I am not diminishing the concerns you bring up but putting them in another perspective. Also consider a real life example as it relates to the Cloud. Most web startups that are creating new jobs are depending on the cloud for their infractructure. Without the Cloud more money would be spent on capital vs salaries if they could even afford to attempt a new business at all.
Sep 28, 2008
Anonymous says:
Very well said. I am sick of hearing about the cloud already. It is just another...Very well said. I am sick of hearing about the cloud already. It is just another attempt to get rid of the people who do all the work so someone else can get a bigger bonus.
Sep 28, 2008
Anonymous says:
Any one know if or when Amazon will offer windows server images ?Any one know if or when Amazon will offer windows server images ?
Oct 02, 2008
Anonymous says:
its coming .. look at http://www.infoworld.com/article/08/10/01/Amazons_EC2_pla...its coming .. look at
http://www.infoworld.com/article/08/10/01/Amazons_EC2_plays_nice_with_Windows_1.html
Sep 29, 2008
Anonymous says:
Chris, Thanks for the analysis. One thing that is missing is any notion o...Chris,
Thanks for the analysis. One thing that is missing is any notion of "elasticity." You are granular at a "weekly" level (40 hours). If the applications are packaged and managed correctly, you could turn them on and off "on-demand." If setup and additional capacity are automated functions, an "elastic" approach yield true variable costs associated with true demand.
Now, I don't want to argue that an hourly level of granularity around system capacity is critical all the time, but it certainly makes sense to shut down what you don't need. Kind of like turning off a light when you leave a room. As long as you are comfortable that getting the light back on is a simple flip of a switch, why not save the energy cost? In this case save all of the costs.
I believe the sentiment on this post generally is correct. Build an internal architecture that enables an elastic approach to application delivery through a seamless "scale out" capability to a cloud provider. I think the foundation for this approach is hypervisor technology coupled with application packaging and lifecycle management (which is what we do at rPath) and data federation technology to share data across datacenters. Good post. Keep them coming.
Billy Marshall, rPath
Oct 01, 2008
Michael Keen says:
Billy you are right on with your reply. I'm working on this very model ri...Billy
you are right on with your reply. I'm working on this very model right now (elasticity in an application delivery and virtual infrastructure to ensure that servers are available and provisioned "on demand" and powered down when not needed. I think it's this model that will make the biggest impact around the adoption of the "cloud" computing model along with DR to the cloud.
Great post.
Cheers
Michael
Oct 03, 2008
Scott Novack says:
Very nice post, Chris. The "premise plus cloud" idea is spot-on, IMHO, and...Very nice post, Chris. The "premise plus cloud" idea is spot-on, IMHO, and I'm just thinking about security and privacy issues.
So, when is Part 2?? I await with baited breath.... (I'm in the Google Group, too, but it's a challenge to keep up with their volume)
Dec 07, 2008
Anonymous says:
Thank you for the calculation. I am currently working on a similar approach to u...Thank you for the calculation. I am currently working on a similar approach to understand the value and costs of cloud technology (TCO,TVO).
I just wanted to hint at a small typo: in the "Cloud" table it should say 24x5x365 instead of 24x7x365. The calculation is correct, though.
Imho the most challenging part in such a comparison is to find a good benchmark (metric) that makes two infrastructure solutions comparable.
Cheers
Markus Klems
http://www.cloudytimes.com
Dec 07, 2008
Chris Fleck says:
Markus, I agree it is difficult to find and provide benchmarks. There are l...Markus, I agree it is difficult to find and provide benchmarks. There are lots of opinions and discussions but hard numbers are scarce. Please share what you have calculated. Thanks for pointing out the typo ( now corrected ).
Nice blog at cloudytimes.com
Apr 05, 2009
Anonymous says:
Why are you using 24x7x365x5 for Amazon EC2. It should be 24x365x5. You are unne...Why are you using 24x
7x365x5 for Amazon EC2. It should be 24x365x5. You are unnecessarily multiplying it by 7.Apr 05, 2009
Chris Fleck says:
the 7 was a typo , not used in the calculation. Now removed.the
7was a typo , not used in the calculation. Now removed.Aug 05
Daniel Smith says:
I have finally got around to adding some of my cloud computing economics researc...I have finally got around to adding some of my cloud computing economics research in Wikipedia's cloud computing article some of them related to domain registration and dedicated server . I tried to maintain a neutral point of view but there's not really a bad thing to say about cloud computing when it comes to economics. I also realised we hadn't yet talked about cloud computing with the oh-so-common electricity utility analogy and as Wikipedia likes analogies for technical content this addition is probably long overdue.
I'm sure the diagram in particular will prove useful and as such I've made it available under a Creative Commons Attribution ShareAlike 3.0 license at the Wikimedia Commons: Cloud computing economics.svg. As it's in Scalable Vector Graphics (SVG) format you can scale it up to poster size with no loss in quality and most browsers support it natively now. The MediaWiki software can also render a PNG version for you if you prefer. As usual it was created using the excellent OmniGraffle software, but I use open source Inkscape and Adobe Illustrator sometimes too (all of which export to SVG).
Without further ado, here's the new section (but it may look quite different by the time you get to it... here's a link to the current revision.
Aug 25
Anonymous says:
I really don't think this is a fair comparison. First of all, for on-premi...I really don't think this is a fair comparison. First of all, for on-premise computing you better always have extra capacity to be able to handle peaks and customer acquisition. On Amazon, you can automatically scale up and down as needed. So the odds of needing the same amount of horsepower in the cloud at all times are low. You also need to factor in the reserved instance pricing where you can now get that same 80 cent instance for 24 cents (plus a minimal one time fee). You didn't factor in any labor costs which are far greater on-premise (can have a new EC2 instance up an running in two minutes), speed to market (how long does it take a normal IT organization to procure a 20 way server?), and many other intangibles. The other issue I have is that after 3-5 years you have an outdated servers that you might decide to replace where in the cloud you can move to larger or more instances.
Another factor that nobody talks about is research and development. My team is building an extreme transaction processing system in the cloud. We are able to simulate huge payloads that would not even be feasible on-premise. We have run tests that fire up 15 XL instances for a few hours trying to find our systems bottlenecks. For our initial launch we don't need half of that horsepower. Experiments like these cost us a couple hundred bucks. In the on-premise world, you wouldn't dream of purchasing that much extra capacity for R&D. Besides, it would take me longer to fill out the paper work then it would to perform the test in the cloud!
My two cents - Mike Kavis www.kavistechnology.com
Aug 25
Chris Fleck says:
All valid points Mike. Time to market or dev/test result is a huge benafit of th...All valid points Mike. Time to market or dev/test result is a huge benafit of the Cloud. There are labor cost savings as well but it depends on the situation. Have you been able to quantify that in a generic way ?
My Part 3 of Cloud Economics does add in the Reserved pricing benefit, although Amazon just reduced that cost further.
Aug 25
Anonymous says:
My scenario is much different. I am in a startup so the cloud vs. on-premi...My scenario is much different. I am in a startup so the cloud vs. on-premise is a no brainer. Our solution requires multiple data centers across the globe. We would not even be in business if we had to raise the money required for on-premise computing. Imagine the cost comparison of building 3-4 real data centers from scratch vs. 3-4 virtual data centers. Like I said, a no-brainer.
I have not quantified the labor costs. Since I am not replacing legacy, I did not need to produce the labor numbers.
Mike Kavis www.kavistechnology.com (too lazy to create an ID)
Oct 01
Anonymous says:
Good article and good insight, putting the whole thing into economic perspective...Good article and good insight, putting the whole thing into economic perspective. However, as a journalist, you really should know the difference between "premise" and "premises"
prem- iseplay_w2("P0527000") (prms)
n. also prem- iss (prms) 1. A proposition upon which an argument is based or from which a conclusion is drawn.
2. Logic a. One of the propositions in a deductive argument.
b. Either the major or the minor proposition of a syllogism, from which the conclusion is drawn.
3. premises Law The preliminary or explanatory statements or facts of a document, as in a deed.
4. premises a. Land and the buildings on it.
b. A building or part of a building.
v. prem- ised, prem- is- ing, prem- is- es
v.tr. 1. To state in advance as an introduction or explanation.
2. To state or assume as a proposition in an argument.
v.intr. To make a premise.
Add Comment