As-a-service models offer huge opportunities, but also complicate security.
Sometimes the easiest way to migrate to a new architectural modelis to let others do the work, others who are experts in their field. This has given rise to many as-a-service models throughout the industry and across the entire technology stack, from software to infrastructure. While this has unlocked huge opportunities to accelerate the deployment of new capabilities or increase economic efficiencies within an organization, it has complicated and even compromised security.
Let’s take a closer look at this trend and the implications. A private cloud is nothing more than the virtualized components of a traditional data center, making it easier to provision, operate, and manage resources more efficiently. Hybrid clouds leverage larger scale public cloud environments to drive further efficiencies. Containers take this a step further, delivering greater micro-segmentation and isolation capabilities with much faster boot times.
In this new reality, the traditional perimeter security model is insufficient. How do you define a perimeter in an environment where any device goes through many networks to many services, both inside and outside the business, or many containers are operating in a single machine?
Security in this any-to-any world must become more dynamic. This means creating an abstract of security functions, like a hypervisor does for operating functions. Security becomes a shared virtual service, applied to workloads and flows instead of machines and physical networks. Automated controls deploy security instances according to policy, reducing the cost and time of deploying new applications or services and taking advantage of the value proposition of virtualization and private/hybrid clouds. Hardware-level security functions help boost performance of the virtual environments while restricting opportunities for leaks. Like software, storage, and other parts of the stack before it, security becomes virtualized, with the same benefits and characteristics.
A Partnered Approach
This new approach to securing enterprise clouds is based on virtual isolation and micro-segmentation. By partnering with the leading virtualization companies, security vendors make sure that each workload deploys dynamically and automatically with virtual sensors that observe and report to the security manager, significantly increasing visibility and control over the virtual environment. Virtual perimeters surround each workload, separating them from each other and from the escalated privileges of the hypervisor. Security policies are linked to the workload, so if a virtual machine or container moves, suspends, or restarts, its policies stay with it. Workloads with different security levels are isolated whether they are on the same physical server or in different data centers, in a virtual machine or a container, reducing the risk of attacks based on privilege escalation or vulnerabilities in the hypervisor.
Security, of course, needs more than just perimeter defenses. The virtual security controller steers traffic to security engines for intrusion detection and prevention, deep file analysis and file reputation management, behavioral analysis, advanced threat defense, and bot detection as needed. Scaling security capacity simply means increasing the number or capacity of these security engines, reducing potential bottlenecks.
One key advantage to this cloud security model is that it applies between layers as well as between workloads. Whether data is flowing in and out of a data-center (north-south), from server to user, or within a data center (east-west) from server to server, it is protected and evaluated by the network gateway, data loss prevention, and other components as defined in your policies.
Consistent policies, protections, and enforcement across your virtual infrastructure are now a reality, as the agility, ubiquity, and efficiency of software-defined security joins the rest of the software-defined infrastructure. This is true cloud security.
View the original post on Dark Reading.