What is the Cloud?
By Avi Deitcher
Cloud seems to be the biggest buzzword in the last few years. Every technology provider, every services provider, if they aren't natively "in the cloud", they are providing a version of their offering "in the cloud."
Although the term "cloud" seems pretty clear to marketers - personally, I am convinced many believe it means, "we can charge more for this if we slap the word 'Cloud' on it" - the majority of people with whom I speak, from engineers and support staff through executives, CEOs and especially customers, do not have a real understanding of what the cloud is, and why it matters.
In order to address these questions, and provide some deeper insight into how to market cloud services, as well as how to evaluate a "cloud provider" by customers, we will be issuing a series of articles on "the Cloud." This is the first in the series.
What is the Cloud?
Wikipedia does a pretty good job providing a concise definition, although it does so in its overview, rather than summary:
"[cloud computing...] relies on sharing of resources to achieve coherence and economies of scale, similar to a utility (like the electricity grid) over a network"
When a company buys electricity, it has two choices:
- Install a power generator.
- Connect to the electricity grid and consume the electricity it needs.
Just about every home and business in the developed world uses the second, or utility model. The expense and effort of maintaining an independent generator, not to mention finding good sources of fuel, are enormous compared to the benefit. So we outsource power generation to a utility. The utility, in turn, generates electricity not just for us, but with a shared set of generators for all of its customers. We then pay just for the electricity we use.
A few key features make this model very appealing:
- We don't know and don't care how the electricity is generated: diesel, coal, nuclear, solar, hydro, etc. As long as the electricity shows up reliably to our mains, we just don't care.
- We don't know and don't care how much excess capacity the utility has or does not have, where it comes from, or how efficient they are in managing their capital expenses. As long as we can pay them a metered rate for what we use or reserver, we just don't care.
You may recognize this as replacing capital expenses, or capex, with operating expenses, or opex. But it is more than just replacing capex with opex; it is replacing all of the overhead and expertise required to manage the capital equipment, as well as the risk of over- or under-investment, with a reliable, outsourced, pay-as-you-go model.
This is precisely the model that cloud provides.
- We don't know and don't care how the services are generated: HP, Sun, Dell servers; Cisco or Huawei network gear; CheckPoint or Cisco firewalls; Java or Ruby on Rails or Node or Python programming language; custom code or off-the-shelf software; etc. As long as the service we want is available from our browser or desktop, we just don't care.
- We don't know and don't care how much capacity the provider has, or how efficient they are in managing their capital expenses. As long as we can pay them a metered rate for what we use or reserve, we just don't care.
As with power, so with computing. We replace capex with opex, but it is more than just that. We replace all of the significant expertise required to maintain servers, operating systems, databases, networks, and software, as well as the risk of over- or under-investment, with a reliable, outsourced, pay-as-you-go model.
Where did it come from?
The concept of "utility computing" has a history almost as long as computing itself. The earliest mainframes were, indeed, timesharing machines for multiple users. However, modern cloud concepts have certain reasonably well-defined characteristics that determine how truly "cloud-y" something is. The level of "cloud-iness" a service has can have a significant impact on a company's agility, speed and profitability.
The nomenclature of "cloud" comes from a common convention used in network and software diagrams. When engineers and architects want to show a large group of computing resources, without showing each one individually because each one does not matter, while the aggregate does, they use a cloud image.
The most common use case is for a large internal network or the Internet itself. The Internet is a vast expense of "things" with which we may or may not interact. We cannot ignore it, but the details of each item are unimportant, hence, it is a "cloud". Thus, the following is a simple diagram of your 2 servers, connected to a firewall, then the Internet:
So what makes a particular service truly cloud vs. not? More importantly, why does it matter to the company providing it, and to the customer? How does a company become a true cloud provider?
We will address all of these in following articles over the coming days.