What Cloud Computing Means to Your Network
As organizations seek to reduce costs, improve efficiency, and increase scalability, virtualization and cloud computing are becoming an integral part of their IT strategies. However, cloud computing also presents new challenges for testers and network/data center operators. Because applications and networks are no longer independent entities, these groups need to devise an efficient process for testing applications, the network, and network devices in the cloud.
Shedding the Silo Approach to Data Center Management
Until recently, many data centers still functioned as silos, with infrastructure, network, applications (end-to-end and services), and operations teams in separate camps. However, as more organizations adopt and experience the advantages of cloud computing and converged applications, they are realizing that the silo model is no longer viable.
Because the network and applications have become so intertwined in a wireless world, it’s imperative (from both a technical and business standpoint) that these two groups work together and not independently. No longer can these teams say: I’m going to design the infrastructure, you design the network, and we’ll layer the software on it. They need to figure out exactly how applications, operations, and network components will work together in a converged network—before they start building anything.
In addition, these teams need to be integrated at both the pre-deployment and post-deployment stages. After all, even if the cloud is running, the organization still might need a new network patch or routing schema. Or, perhaps they need to update or deploy a new service. As a result, the lines between pre- and post-deployment are becoming less rigid. And as applications and the network converge, the teams that work on them will need to do the same.
Rethinking Roles and Responsibilities
In addition to dismantling the silo approach to data center management, virtualization is forcing different groups to better define their responsibilities. With cloud computing, a portion of the network is virtualized and can be invisible to network operators. Is that area then still considered part of the data center? To avoid such confusion, organizations need to clearly designate who is responsible for setting up and testing network devices in a virtualized infrastructure—and resolving any issues.
For example, when VMware introduced the virtualized switch, application and operations administrators took over the configuration of these switches. This created problems for the networking team, which usually was responsible for managing and configuring switches. When the network would go down due to a configuration error in this switch, the networking team had no idea what the cause was. After figuring out this problem, many organizations returned switch management and support responsibilities to the networking team. Nevertheless, solving a problem in a cloud environment sometimes requires assistance from all teams.
A New Strategy for Testing in the Cloud
Similarly, organizations need to take a different approach to testing in the cloud. Many teams know how to test networks or applications in static data environments, but because the cloud is always changing, this adds another layer of complexity. For instance, when a team spins up more servers to power up an application, the server needs to know where the database is and the security privileges—and the firewall needs to allow access. However, servers might not always spin up in the same place, the same IP, or the same hardware location. And today this box is the application server; tomorrow it might be something else.
So, how do you test a network or an application that is constantly changing? Organizations need to think of a converged application as a chain that is only as strong as its weakest link. Each link is like a service, and network and application teams must be able to test every link independently to ensure that it’s as strong as it needs to be.
The ideal testing strategy for a cloud environment therefore combines traditional end-to-end testing—or black box testing—with white box and gray box testing. After conducting black box testing for multiple scenarios (in which boxes have been spun up or down), organizations can use white box and gray box testing to test each service, application, and network link independently.
By individually testing each link, network and application teams can be confident that as the cloud continues to morph, the converged application will still meet business requirements and deliver the necessary service levels.