CLOUD COMPUTING IS THE CLOSEST WE'VE COME yet to what futurists promised the internet could be. If you use Google Apps, YouTube, Amazon, Salesforce, Flickr, Facebook, Bittorrent, Skype, or any of a myriad of applications that let you access and share information quickly and connect with other people, you’ve already been in the cloud. Below are 10 questions and answers to get you up to speed on this technology.
In simple terms, cloud computing is a virtual network of services and infrastructure that can be accessed from anywhere at anytime. But at the same time, there is no “thing” called the cloud; it’s a general term applied to this distributed computing network.
Although they share some features, grids and thin clients are localized, while clouds represent something more, says Greg Arnette, founder and CTO of Sonian Networks, a hosted services company that helps clients manage critical IT functions.
“We’re entering a period of time where hybrid architectures are going to be in place for the foreseeable future. The hybrid architecture element means that there will be a blending of IT services that are on-premises—the core servers behind the firewall—and there will be extension services running in the cloud, tightly tethered to the customer’s on-site data center,” he says. “Think of the cloud as a virtual extension of the customer’s own data center that happens to be running on someone else’s infrastructure, but it’s a set of services that the customer has complete and full access to.”
Done correctly, there are numerous benefits to cloud computing, including powerful (yet inexpensive) server capabilities, software-as-a-service (SaaS), data backup and storage, IT sandboxing capabilities, and more.
The biggest benefit for bottom-line-conscious business offices and IT departments is that you don’t “buy” the cloud. Much like a common utility, you just pay for what you use, when you use it, and then turn it off when you’re done. Think about situations in which your normally smooth-running servers can be inundated with data requests, such as at student registration time. You could invest thousands of dollars in additional servers and staff to handle that load, but if those servers sit unused for most of the time, it’s a waste of money. “The ability to have that server somewhere, to not have to worry about it, turn it up as you need it, and pay for only what you use is really attractive to a lot of people,” says Mike Richwalsky, assistant director of public affairs at Allegheny College (Pa.). Richwalsky also writes HighEdWebTech.com, a blog that deals with web development in higher education.
“Budgets are getting smaller and, at least at schools that I have been at, the information technology group has to be everything for everyone service-wise,” he says. “That’s very difficult to do with budgets that are being cut and with staff that aren’t being replaced. Clouds offer us the opportunity to accomplish the things that we want to accomplish with the resources that we have.”
It can cost less to take advantage of existing infrastructure than to build your own from scratch, especially for short-term projects. Because of that, cloud technology could be the saving grace for business continuity during catastrophic events such as hurricanes, floods, or terror attacks, and could prove invaluable for crisis communications.
Michael Dame, director of web communications at Virginia Tech at the time of the tragic campus shootings in 2007, experienced unexpected demand firsthand. In the furious few days following the event, the school’s servers were inundated with information requests. In fact, during one 24-hour period Virginia Tech served 432 GB of traffic—a capacity that would cripple many smaller school networks. Fortunately, Dame was able to contact a local networking group and request that four additional servers get turned on to help with the heavy demand.
That’s not something most other institutions can easily do, Richwalsky says. “I’m at a smaller institution and I don’t have that ability. We don’t have four spare web servers sitting around and ready to go. But you can accomplish the same thing in the cloud if you have the infrastructure set up. All you need to do is flip a switch. You could have extra servers set up to handle the load and get everyone the information that they need and not have to worry about the infrastructure.”
Richwalsky says that by accessing Amazon’s cloud service he can accomplish what Virginia Tech did for less than $200.
The real economy of scale of clouds comes in its expandability. Most institutions have sufficient server capabilities for normal operations, but there are times when that’s not enough. An example is a distance learning system that runs on a typical server. It operates quite well under normal circumstances. But what happens when, as we’ve seen recently, there is a large increase in distance learning students?
“Eventually that one server is going to run out of capacity,” Richwalsky says. “If you think about the cost of replacing that server, you might spend $10,000 to $15,000 by the time you’re done with software, support, and so on. But if you go into the cloud you don’t have that cost. A small server at Amazon, for example, is only 10 cents an hour, which is about $70 a month if you were to run it every second of every day of the month.”
To keep it safe from prying eyes, many of the cloud providers encrypt user data when it is uploaded. Still, says Richwalsky, the security of a system is only as good as the protection you give it, and institutions should really consider what they use the cloud for.
“There is some data, such as our student registration, that will always reside on campus,” he explains. “If you’re buying a cloud server and you are uploading code and applications, that’s no different than a server that lives in your data center. You are going to put the same protections on it that you put in your on-campus infrastructure. You are going to lock out certain ports, you are going to make sure you are always patching and getting the most up-to-date software, and so on. If your data is not secure on your own campus, it is not going to be any more or less secure in the cloud.”
“It’s essential that you look beyond the technology gains and scrutinize the cloud provider,” suggests Dwayne Melancon, vice president of corporate and business development at Tripwire, a company that provides enterprise-level configuration control software. “Alongside guarantees from the provider, you must also ensure that they have an alternative strategy in place in case of disruptions or loss of connectivity, including awareness of the provider’s fallback plans that may jeopardize valuable information.”
To that end, many cloud providers have begun offering service level agreements, or SLAs, that guarantee a level of connectivity and service and that credit accounts if the servers fail.
Salesforce.com is a widely used CRM application that originated as a cloud application. Google Apps is another example of a cloud SaaS, where the application and the data it creates reside remotely. Microsoft is getting ready to launch its own cloud-based version of the Office productivity suite.
Greg Arnette of Sonian Networks says his company uses the cloud as a digital content archiving service focusing on e-mail, instant messaging, and files. “It gets that information off the customer’s local area network storage footprint and puts it up in the cloud in a protected virtual storage silo that is dedicated only to that particular customer. The cloud is a better place to manage an ever-growing long-term archive that needs to be searched and data-mined extensively very quickly and efficiently,” he says. “An on-premise archiving system just continues to be a drain on the IT staff’s time and budget, because the archive is always getting bigger and needs to be indexed periodically.”
Richwalsky points out that the cloud has a lot of potential in the classroom and the sciences, beyond core services and IT services. Besides serving distance learning content and video, researchers can run complex simulations in the cloud without taxing local servers. “You can do unit testing on a large scale, instead of using clusters,” Richwalsky says. “You can turn on 1,000 servers for an hour, run some major tests, turn them off, and you’ve spent a relatively small amount. It is so much more cost-efficient than building a cluster that might be used three or four times a semester.”
The better question might be “Who isn’t getting in on cloud technology?” Joining the companies listed earlier in this space are such industry leaders as Microsoft, IBM, Dell, AT&T, Sun Microsystems, and a host of smaller companies.