The Economics of Cloud Computing

By George Reese
October 24, 2008 | Comments: 16

Cloud computing has been "the next cool thing" for at least the past 18 months. The current economic climate, however, may be the thing that accelerates the maturity of the technology and drives mainstream adoption in 2009.

This economic crisis is very different from the normal ebbs of the business cycle we have grown accustomed to. The difference lies in how rapidly sources of capital have dried up—whether that capital is coming from venture capitalists, angel investors, or banks. Capital markets are frozen, and companies needing to make capital investments to continue operations or grow are facing a daunting challenge.

A lack of capital creates a lack of flexibility in leveraging technology to operate and grow a business. If you can't get access to a bank loan, you have to use your company revenues to buy new servers. Using company revenues can damage cash flow, harm valuations, and put otherwise healthy businesses at risk.

Typically, when a company wants to grow its IT infrastructure, it has two options:

  • Build it in house and own or lease the equipment.
  • Outsource the infrastructure to a managed services provider.

In both scenarios, a company must purchase the infrastructure to support peak usage regardless of the normal system usage. One Valtira client, for example, has fairly low usage for most of the year, but sees usage equalling millions of page views/month for about 15 minutes each quarter. For both of the above options, they are faced with paying for an infrastructure necessary for only about 1 hour each year when something much more minimalist will support the rest of the year.

Let's assume for this customer that two application servers backed by two database servers and balanced by a load balancer will solve the problem. The options look something like this:

Internal IT Managed Services
Capital Investment $40,000 $0
Setup Costs $10,000 $5,000
Monthly Services $0 $4,000
Monthly Labor $3,200 $0
Cost Over 3 Years $149,000 $129,000

For this calculation, I assume fairly baseline, server-class systems with a good amount of RAM and on-board RAID5 such as a Dell 2950 and a good load balancer. The 3-year cost assumes a 10% cost of capital. I also assume a very cost-efficient managed services provider. Most of the big names will be at least three times more expensive than the numbers I am providing here.

Under this scenario, managed services saves you a nice 13.5% over the do it yourself approach (assuming you don't get taken to the cleaners by one of the big managed services companies). Of course, it does not consider at all the impact of a server outage at 3am, which is where managed services will shine.

What is particularly appealing about managed services, however, is the lack of capital investment. The $40,000 up-front for an internal IT approach is a terrible burden in the current economic environment. Even if you can get credit, the cost of the loan makes that $40,000 much more expensive over three years than $40,000.

Good argument for managed services? Yes, but a better argument for the cloud.

The cloud enters the picture looking like this:

Managed Services The Cloud
Capital Investment $0 $0
Setup Costs $5,000 $1,000
Monthly Services $4,000 $2,400
Monthly Labor $0 $1,000
Cost Over 3 Years $129,000 $106,000

Cloud savings over internal IT jump to 29% without getting into the discussion of buy for capacity versus buy what you use!

Between managed services and the cloud, the cloud provides 18% savings.

While 18% and 29% savings are nothing to sneeze at, they are just the start of the financial benefits of the cloud. It goes on.

  • No matter what your needs, your up-front cost is always $0
  • As the discrepancy between peak usage and standard usage grows, the cost difference between the cloud and other options becomes overwhelming.
  • The cloud option essentially includes a built-in SAN in the form of the Amazon Elastic Block Storage. The internal IT and managed services options would go up significantly if we added the cost of a SAN into the infrastructure.
  • Cheap redundancy! While the above environment is not quite a "high availability" environment, it is very highly redundant with systems spread across multiple data centers. The managed services and internal IT options, on the other hand, have single physical points of failure as the application servers and database servers are likely located in the same rack.

Let's say, however, that you need 10 servers to handle peak usage for 1 hour each year and just 2 to operate the rest of the year. Ignoring the impact of the cost of capital:

  • Internal IT adds another $40,000 in total costs over 3 years.
  • Managed services adds another $144,000 in total costs over 3 years.
  • The Amazon Cloud adds about $24 in total costs over 3 years.

No, that was not a typo. That's forty THOUSAND dollars against one hundred forty-four THOUSAND dollars against 24 dollars. And as I mentioned earlier, this setup is based on an actual Valtira client that was considering a dedicated managed services option before Valtira began deploying customers in the Amazon cloud. It is not some contrived example.

Obviously, most organizations have either seasonal peaks or daily peaks (or both) with a less dramatic cost differential; but the cost differential is still quite dramatic and quite impactful to the bottom line. In addition, the ability to pay for what you use makes it easy to engage in "proofs of concept" and other R&D that requires dedicated hardware.

You might also be interested in:


That's the best financial justification I've seen thus far. Thanks for sharing some real numbers.

I appreciate the insights offered in this article but I would like to share a slightly different perspective.

I believe the financial models are inaccurate in two areas. First, internal IT is going to incur costs for monthly services. They need to buy bandwidth from an Internet provider and pay for power+cooling infrastructure. Or, internal IT needs to colocate the servers in a datacenter. Second, the $4,000 monthly cost for a couple 2950's and a load balancer seems high. The going rate for this type of service from both regional and national hosters is oftentimes less than $2,500/mo.

Price is not the major reason that customers choose internal IT vs. managed hosting vs. cloud. Yes, it is a factor but it is not the primary factor. The biggest factor is whether or not the customer wants to manage their servers. If the customer does not want to maintain and troubleshoot their services then the internal IT and cloud models are not viable options.

Money is everything actually.
Would the decommission cost be considered too?

Nice meeting in person today, George.

Money is everything actually.
Would the decommission cost be considered too?

Nice meeting you in person today, George.
Great presentaion.

Great site.Thanks a lot.

nice,thanks for sharing.

Thank you for your information. söve fiyatları.

Thanks a lot.


Thanks you too. Statistic is very interesting.

Wonderful! All it very nice. Thanks you.

Uff, this is good as good! Thanks you too dude!

What you point out here is a paradox. You claim that Cloud computing is always cheaper than internal IT but a Cloud provider uses internal IT to provide Cloud services !

Could you please elaborate on the monthly labor costs?

It is misleading that the financial picture states the total cost for 3 years, as the calculation only includes 31 months and not 36.

In addition, if you were to do the cost over a 10-year period, internal IT is cheaper than managed services - might not say much, but still worth pointing out!

This is a great article and really helps me understand the economics of managed hosting and cloud. I did have a couple of basic questions though and am probably missing something.

First - if I simply take monthly costs line items for 3 years or 36 months - I get a higher number than the total cost line. For example in managed services - $4,000 per month for 36 months is $144,000 vs the total cost of $129,000 cited in the chart. Same for cloud 3,400 in monthly costs x 36 months = $122,400 vs the $106,000 total cost in the chart. Am I missing something?

Also - I am just curious if you know where I could find an updated version of these estimates given that cloud computing costs have dropped pretty meaningfully since this was published 3 years ago.


News Topics

Recommended for You

Got a Question?