If you can deploy all of your custom-built software systems on cloud hardware and leverage SaaS systems for your packaged software, you might be able to achieve an all-cloud IT infrastructure. Table 1-1 lists the components of the typical small- or medium-sized business.
MS Outlook, Apple Mail
Gmail, Yahoo!, MSN
SAP CRM/Oracle CRM/Siebel
Microsoft Office/Lotus Notes
Server, racks, and firewall
Amazon EC2, GoGrid, Mosso
The potential impact of the cloud is significant. For some organizations—particularly small- to medium-sized businesses—it makes it possible to never again purchase a server or own any software licenses. In other words, all of these worries diminish greatly or disappear altogether:
When do I schedule my next software upgrade? SaaS vendors perform the upgrades for you; you rarely even know what version you are using.
What do I do when a piece of hardware fails at 3 a.m.? Cloud infrastructure management tools are capable of automating even the most traumatic disaster recovery policies.
How do I manage my technology assets? When you are in the cloud, you have fewer technology assets (computers, printers, etc.) to manage and track.
What do I do with my old hardware? You don’t own the hardware, so you don’t have to dispose of it.
How do I manage the depreciation of my IT assets? Your costs are based on usage and thus don’t involve depreciable expenses.
When can I afford to add capacity to my infrastructure? In the cloud, you can add capacity discretely as the business needs it.
SaaS vendors (whom I’ve included as part of cloud computing) can run all their services in a hardware cloud provided by another vendor, and therefore offer a robust cloud infrastructure to their customers without owning their own hardware. In fact, my own business runs that way.
Internal IT infrastructure and support
Outsourcing to managed services
If you own the boxes, you have an internally managed IT infrastructure—even if they are sitting in a rack in someone else’s data center. For you, the key potential benefit of cloud computing (certainly financially) is the lack of capital investment required to leverage it.
Internal IT infrastructure and support is one in which you own the boxes and pay people—whether staff or contract employees—to maintain those boxes. When a box fails, you incur that cost, and you have no replacement absent a cold spare that you own.
Managed services outsourcing has similar benefits to the cloud in that you pay a fixed fee for someone else to own your servers and make sure they stay up. If a server goes down, it is the managed services company who has to worry about replacing it immediately (or within whatever terms have been defined in your service-level agreement). They provide the expertise to make sure the servers are fixed with the proper operating system patches and manage the network infrastructure in which the servers operate.
Table 1-2 provides a comparison between internal IT, managed services, and cloud-based IT with respect to various facets of IT infrastructure development.
How much cash do you have to cough up in order to set up your infrastructure or make changes to it? With internal IT, you have to pay for your hardware before you need it (financing is not important in this equation).[a] Under managed services, you are typically required to pay a moderate setup fee. In the cloud, you generally have no up-front costs and no commitment.
Based on usage
Your ongoing costs for internal IT are based on the cost of staff and/or contractors to manage the infrastructure, as well as space at your hosting provider and/or real estate and utilities costs. You can see significant variances in the ongoing costs—especially with contract resources—as emergencies occur and other issues arise. Although managed services are often quite pricey, you generally know exactly what you are going to pay each month and it rarely varies. The cloud, on the other hand, can be either pricey or cheap, depending on your needs. Its key advantage is that you pay for exactly what you use and nothing more. Your staff costs are greater than with a managed services provider, but less than with internal IT.
How long does it take to add a new component into your infrastructure? Under both the internal IT and managed services models, you need to plan ahead of time, place an order, wait for the component to arrive, and then set it up in the data center. The wait is typically significantly shorter with a managed services provider, since they make purchases ahead of time in bulk. Under the cloud, however, you can have a new “server” operational within minutes of deciding you want it.
How easily can your infrastructure adapt to unexpected peaks in resource demands? For example, do you have a limit on disk space? What happens if you suddenly approach that limit? Internal IT has a very fixed capacity and can meet increased resource demands only through further capital investment. A managed services provider, on the other hand, usually can offer temporary capacity relief by uncapping your bandwidth, giving you short-term access to alternative storage options, and so on. The cloud, however, can be set up to automatically add capacity into your infrastructure as needed, and to let go of that capacity when it is no longer required.
Staff expertise requirements
How much expertise do you need in-house to support your environments? With internal IT, you obviously need staff or contractors who know the ins and outs of your infrastructure, from opening the boxes up and fiddling with the hardware to making sure the operating systems are up-to-date with the latest patches. The advantage here goes to the managed services infrastructure, which enables you to be largely ignorant of all things IT. Finally, the cloud may require a lot of skill or very little skill, depending on how you are using it. You can often find a cloud infrastructure manager (enStratus or RightScale, for example) to manage the environment, but you still must have the skills to set up your machine images.
Moderate to high
How certain are you that your services will stay up 24/7? The ability to create a high-availability infrastructure with an internal IT staff is a function of the skill level of your staff and the amount of cash you invest in the infrastructure. A managed services provider is the safest, most proven alternative, but this option can lack the locational redundancy of the cloud. A cloud infrastructure, finally, has significant locational redundancies but lacks a proven track record of stability.
[a] From a financial perspective, the difference between coughing up cash today and borrowing it from a bank is inconsequential. Either way, spending $40K costs you money. If you borrow it, you pay interest. If you take it out of your bank account, you lose the opportunity to do something else with it (cost of capital).
The one obvious fact that should jump out of this chart is that building an IT infrastructure from scratch no longer makes any sense. The only companies that should have an internal IT are organizations with a significant preexisting investment in internal IT or with regulatory requirements that prevent data storage in third-party environments.
Everyone else should be using a managed services provider or the cloud.
Perhaps the biggest benefit of cloud computing over building out your own IT infrastructure has nothing to do with technology—it’s financial. The “pay for what you use” model of cloud computing is significantly cheaper for a company than the “pay for everything up front” model of internal IT.
The primary financial problem with an internally based IT infrastructure is the capital cost. A capital cost is cash you pay for assets prior to their entering into operations. If you buy a server, that purchase is a capital cost because you pay for it all up front, and then you realize its benefits (in other words, you use it) over the course of 2–3 years.
Let’s look at the example of a $5,000 computer that costs $2,000 to set up. The $5,000 is a capital cost and the $2,000 is a one-time expense. From an accounting perspective, the $5,000 cost is just a “funny money” transaction, in that $5,000 is moved from one asset account (your bank account) into another asset account (your fixed assets account). The $2,000, on the other hand, is a real expense that offsets your profits.
The server is what is called a depreciable asset. As it is used, the server is depreciated in accordance with how much it has been used. In other words, the server’s value to the company is reduced each month it is in use until it is worth nothing and removed from service. Each reduction in value is considered an expense that offsets the company’s profits.
Finance managers hate capital costs for a variety of reasons. In fact, they hate any expenses that are not tied directly to the current operation of the company. The core rationale for this dislike is that you are losing cash today for a benefit that you will receive slowly over time (technically, over the course of the depreciation of the server). Any business owner or executive wants to focus the organization’s cash on things that benefit them today. This concern is most acute with the small- and medium-sized business that may not have an easy time walking into the bank and asking for a loan.
The key problem with this delayed realization of value is that money costs money. A company will often fund their operational costs through revenues and pay for capital expenses through loans. If you can grow the company faster than the cost of money, you win. If you cannot grow that rapidly or—worse—you cannot get access to credit, the capital expenses become a significant drain on the organization.
Managed services infrastructures and the cloud are so attractive to companies because they largely eliminate capital investment and other up-front costs. The cloud has the added advantage of tying your costs to exactly what you are using, meaning that you can often connect IT costs to revenue instead of treating them as overhead.
Table 1-3 compares the costs of setting up an infrastructure to support a single “moderately high availability” transactional web application with a load balancer, two application servers, and two database servers. I took typical costs at the time of writing, October 2008.
Monthly service fees
Monthly staff costs
Net cost over three years
Table 1-3 makes the following assumptions:
The use of fairly standard 1u server systems, such as a Dell 2950 and the high-end Amazon instances.
The use of a hardware load balancer in the internal IT and managed services configuration and a software load balancer in the cloud.
No significant data storage or bandwidth needs (different bandwidth or storage needs can have a significant impact on this calculation).
The low end of the cost spectrum for each of the options (in particular, some managed services providers will charge up to three times the costs listed in the table for the same infrastructure).
Net costs denominated in today’s dollars (in other words, don’t worry about inflation).
A cost of capital of 10% (cost of capital is what you could have done with all of the up-front cash instead of sinking it into a server and setup fees—basically the money’s interest rate plus opportunity costs).
The use of third-party cloud management tools such as enStratus or RightScale, incorporated into the cloud costs.
Staff costs representing a fraction of an individual (this isolated infrastructure does not demand a full-time employee under any model).
Perhaps the most controversial element of this analysis is what might appear to be an “apples versus oranges” comparison on the load balancer costs. The reality is that this architecture doesn’t really require a hardware load balancer except for extremely high-volume websites. So you likely could get away with a software load balancer in all three options.
A software load balancer, however, is very problematic in both the internal IT and managed services infrastructures for a couple of reasons:
A normal server is much more likely to fail than a hardware load balancer. Because it is much harder to replace a server in the internal IT and managed services scenarios, the loss of that software load balancer is simply unacceptable in those two scenarios, whereas it would go unnoticed in the cloud scenario.
If you are investing in actual hardware, you may want a load balancer that will grow with your IT needs. A hardware load balancer is much more capable of doing that than a software load balancer. In the cloud, however, you can cheaply add dedicated software load balancers, so it becomes a nonissue.
In addition, some cloud providers (GoGrid, for example) include free hardware load balancing, which makes the entire software versus hardware discussion moot. Furthermore, Amazon is scheduled to offer its own load-balancing solution at some point in 2009. Nevertheless, if you don’t buy into my rationale for comparing the hardware load balancers against the software load balancers, here is the comparison using all software load balancers: $134K for internal IT, $92K for managed services, and $106K for a cloud environment.
If we exclude sunk costs, the right managed services option and cloud computing are always financially more attractive than managing your own IT. Across all financial metrics—capital requirements, total cost of ownership, complexity of costs—internal IT is always the odd man out.
As your infrastructure becomes more complex, determining whether a managed services infrastructure, a mixed infrastructure, or a cloud infrastructure makes more economic sense becomes significantly more complex.
If you have an application that you know has to be available 24/7/365, and even 1 minute of downtime in a year is entirely unacceptable, you almost certainly want to opt for a managed services environment and not concern yourself too much with the cost differences (they may even favor the managed services provider in that scenario).
On the other hand, if you want to get high-availability on the cheap, and 99.995% is good enough, you can’t beat the cloud.