The idea of cloud computing — designed around an architecture whose natural state is a shared pool outside the enterprise — has gained momentum in recent months as a way to reduce cost and improve IT flexibility. But the use of cloud computing also carries with it security risks, including perils related to compliance, availability and data integrity.
Mmany companies don't think through those risks upfront. For example, having proper failover technology in place is a component of securing the cloud that is often overlooked, notes Josh Greenbaum, principal at Enterprise Applications Consulting. Yet everyone makes sure they have failover for established services, like electricity. "If you look around, go to any major facility, what is sitting in a box outside is an alternative power supply. They don't rely on just the grid," says Greenbaum. He argues that cloud computing should be no different.
In some cases, the risk is too great to rely on the cloud. And where the decision is made to put some services and applications in the cloud, the business must ask how that risk should be managed.
David Cearley, a vice president and fellow at Gartner, says placing limits on the use of cloud technology is a subtle issue that companies have to examine closely, measuring the risk against when and where cloud computing can be effective. For example, by giving up some control over the data, companies get in exchange cost economies. IT, along with other C-level executives, must decide if that trade-off is worthwhile. Cearley says that everything will eventually be available as a cloud service — but at any individual business, not everything will be accessed from the cloud.
"In a shared pool outside the enterprise, you don't have any knowledge or control of where the resources run. So if you have a concern over data location, as an example, that may be a reason for not using it," Cearley says.
There is a huge body of standards, including services like SAS Interaction Management, for example, that apply for IT security and compliance, governing most business interactions that will, over time, have to be translated to the cloud, notes Greenbaum.
But in the meantime, until security models and standards emerge for cloud computing architecture, most of the risk and blame if something goes wrong will fall directly on the shoulders of IT — and not on the cloud computing service providers. "The Salesforce.coms and NetSuites of the world don't offer the kind of governance, risk, and compliance [mechanisms] mandated by regulatory regimes," Greenbaum says.
Ultimately, the consumer of the services is responsible for maintaining the confidentiality, integrity and availability of data, agrees Kristin Lovejoy, director of IBM's security, governance and risk management division.
As far as placing limitations on when to deploy the cloud, Lovejoy advises that companies adhere to Geoffrey Moore's consideration of "context versus core." (Moore is a business strategist and managing partner of consultancy TCG Advisors).
According to Moore, core business practices provide competitive differentiation. Context practices deliver business activities that are typically internal, such as HR services and payroll. Both core and context can be divided into mission-critical applications and non-mission-critical ones. "If a non-mission-critical application goes offline, the company can survive," Lovejoy says.
The rule of thumb Moore comes up with, notes Lovejoy, is this: If the business practice is context and non-mission-critical, then always put it in the cloud. If it is context and mission-critical, it is likely you should make it cloud-enabled. However, if it is core and non-mission-critical, you may want to think about keeping it behind the firewall; if it is core and mission-critical, then definitely keep it behind the firewall, she says.
The cloud approach doesn't map naturally to how good security is typically designed, says John Pescatore, Gartner's chief security analyst.
The area that worries Pescatore most is how quickly cloud-based services are updated and changed. He cites Microsoft's painstaking development of the SDLC (Software Development Life Cycle) initiative that assumes mission-critical software will have a three- to five-year period in which it will not substantially change.
"In the cloud, every two weeks we add a new feature, changing the app all the time. But the secure SDLC is not built to do that. We are going back to the old Netscape days of pushing out new features real quick, and nobody has a security cycle that moves that fast," Pescatore says.
What makes matters even worse is that the business user can't say he wants to stay on the old version. "In the cloud you have to accept the next version, possibly nullifying any security that was built into the old application or assumed through integration at the customer site.
In the near term, where security is core and mission-critical — at the upper end of the enterprise, at financial institutions, and in government — IT will have to add its own security layer. That means cloud computing won't be any cheaper than running the application in-house, says Pescatore.
"But a few years from now, strong security will be pushed down a layer, and there will be enough security baked in for all but perhaps the DoD," he says.
But if larger enterprises can't yet rely on security in the cloud-provided security, smaller companies may actually get better security from the cloud, says Pescatore. One reason is that a cloud provider can invest more in security than any individual small business could, because the cost is applied across hundreds of customers. Another is that as soon as a cloud provider patches a security vulnerability, all its customers are protected immediately, unlike the case for downloadable patches that IT must apply itself.
Another issue that can affect customers large and small — the location of their data — will see a shift in the next few years, Pescatore says. That's especially important for companies that do business across national boundaries, as different privacy and data management laws apply in different countries.
Today, you never know where your data is stored in the cloud. And that fact raises all sorts of compliance issues around data privacy, segregation and security. But this indeterminate location is beginning to change. For example, Google lets customers specify where their Google Apps data is stored, thanks to its acquisition of Postini, an email security company. For example, a Swiss Bank may wants its customer data files stored in Switzerland, which Google can now do.
Further out is the ability to physically separate your data from other customers' data in the cloud's multitenant architecture, says Pescatore. But he foresees such separation being enabled from the still nascent but increasingly powerful virtualisation technology.
The security experts agree that no matter how stringent a vendor's SLA (service-level agreement) or how strong its security technology, reducing risk in any environment rests with the individual company's willingness to audit what is and what is not working.
Enterprise Consulting Group's Greenbaum says we all have a bit of the "control freak" in us, trusting no one but ourselves. Perhaps that is a good thing, he says. When it comes to cloud computing, the experts fundamentally offer the same advice, which can be summarised by two very famous quotations, one an old Russian proverb that President Reagan liked to use -- "trust but verify" -- and the other by Intel's famous former CEO, Andy Grove -- "only the paranoid survive".