Cloud security is in a state of flux. Actually it’s not really, but discussion is rife across the industry regarding the general state of cloud data and application welfare. Questions flutter surrounding whether particular services are secure, whether multi-tenant clouds can be locked down and whether “spun down” instances of cloud servers will remain secure so that old customer data is deleted and non-accessible.
So where are we headed with the general consensus on cloud security just now?
A cloud is just a server
Of course we should start out from first principles and look to comments made by the big name cloud providers who have pointed out that the cloud is only as secure as the controls you place upon it i.e. virtualised hosting services do not come with “security included” as some sort of free gift... and so a cloud server, is really just a server, but delivered as a service.
So how do we now move to mitigate risks in any given cloud deployment scenario and at what point do we bring the shielding defences to bear upon the data that we seek to protect? Should we look to endpoint security controls as a primary security consideration, or should we focus on network back doors themselves, or both?
Companies such as F5 Networks appear to already be working at the IP address layer to provide more tailored network architectures with better “resource visibility” as applications need to deal with the daily duress of real-world data flows.
But are IP intelligence and application delivery controls enough? These technologies seek to champion the “automation” of application protection mechanisms where they may be enforced. Vendors in this space are also fond of talking about “contextual awareness” i.e. not just for applications so that they can gauge user requirements more intuitively, but also for cloud network traffic controls that can identify and filter data via a Content Delivery Network (CDN) if it appears to be malicious.
The basic facts are...
In basic terms, we have a multiplicity of application controls, data transport vectors and network gateways to deal with in a global cloud computing environment that is barely sure enough of itself to settle upon one single agreed set of open architectural standards.
Julian Lovelock of identity assurance and authentication specialist company ActivIdentity sees four “roads” for data travel down in the cloud. Each of these virtual crossroad paths may give us some clues as to the best route to deliver an Identity and Access Management (IAM) solution he says.
1) Open Access: data is accessible on the public Internet where usernames and passwords are managed by SaaS providers -- this solution is argued to offer the “least amount of protection for your data” with minimal control.
2) Behind the VPN: this option enables remote users to first authenticate to the VPN (most likely via a One Time Password (OTP) solution), then enter username and password.
3) Federated Identity Management: the user authenticates to a central portal through which they gain access to multiple applications.
4) Native Strong Authentication: strong authentication deployed separately in each, individual cloud software application.
ActivIdentity’s laying down of these four crossroad paths is unlikely to be an exhaustive future-proofed definition of the complete who, when, why, what and where of how we need to approach this issue, but its arguably a pretty good starting point in an arena largely bereft of clear definitive statements in this vein.
We must also look to the multifarious nature of threats that can develop here from Advanced Persistent Threats (APTs), to ad hoc hacking and onward to attacks that may originate from internal or former employees. While security vendors in this space would like us to believe that solutions do exist that do not compromise upon usability or convenience in the face of the Bring Your Own Device (BYOD) phenomenon, to place complete trust in any perimeter defence at this stage still requires what appears to be a considerable leap of blind faith.
Are we being too sceptical and cynical here by far? HP’s approach to cloud security appears to be somewhat more open ended than some; the firm’s central cloud security portal says upfront that it exists to try and “determine future cloud computing requirements” as well as data centre and application design principles.
Open research has the answer
The answer to our worries and concerns in this space will most likely not come from any single vendor or solution provider of any kind. The real answers many purists would argue are most likely to come from lab research collaborations such as the Gnosis process simulation language and tool that seek nothing more than the truth without and brand-driven agenda.
So, will cloud security provisioning and protection keep pace with the general state of cloud data and application hosting over the next five years? The answer is that is has to -- otherwise we will fail to grasp the wider and deeper benefits of virtualisation as they exist in full.