Network Visibility in Hybrid Multi-Cloud Environments

A sales representative at a remote branch office just opened a ticket with support. She is unable to load new emails as well as the team group chat on her laptop when connected to the corporate WiFi. As a result, she had to tether her laptop to her mobile phone and use her own personal data plan to fuel her work. A developer at the main building just stopped by your desk and mentioned that uploading a large file to a remote cloud server is taking longer than usual. The Director of IT just sent you an email that she is hardly able to capture the conversation on the weekly management web-conference meeting. How do network engineering teams deal with these problems? By detecting problems before they are escalated, and solving them as quickly as possible…

Network visibility is a requirement to successfully manage a hybrid multi-cloud environment.

Today’s network administrators are constantly engaged with researching, designing, building, and upgrading a part of the network. At the beginning of a network design project, performance monitoring isn’t necessarily a priority. However, sooner or later that network goes live and starts servicing end users. Then, performance monitoring becomes a real need to gain network visibility and successfully operate the network. Having a system in place that gathers network performance metrics is necessary to achieve operational efficiencies and minimize service interruptions and support tickets. If the network offers connectivity and hosting services to customers, then consider performance monitoring a business requirement to assure quality and meet service level agreements.

Network visibility in hybrid multi-cloud environments requires the implementation of a distributed, active, continuous, and end-to-end network performance monitoring solution.

Hybrid multi-cloud networks extend across a large geographical area, sometimes spanning different countries or even continents. A distributed performance monitoring solution should run on top of monitoring agents that support a variety of networks such as WAN, WiFi, cloud, and virtual. This capability ensures that the measurements are executed across the entirety of the network, and everywhere in which the end users are located. The distributed monitoring sensors run active and continuous tests against remote targets, which include remote network nodes, services, or SaaS applications.

Active monitoring tests will report metrics such as network latency, packet loss, and application response time. The continuous monitoring is required to build a performance baseline to proactively detect service interruptions and performance degradations.

Lastly, end-to-end measurements ensure that communication between endpoints, such as client-client, client-server or server-server, are granted by the network.

If we take as reference the cloud observability maturity model presented at ONUG Europe 2019, network visibility will sit in the prevention layer.

Network visibility is achieved by simulating user experience with active end-to-end traffic against network services and applications. This level of monitoring will assure minimal downtime via quick detection of network and application issues. So where we do we go from here? Before closing, I’d like to have one last word about the importance of a smooth deployment and efficient maintenance of a distributed network performance monitoring tool, at scale.

Linux containers simplifies deployment in large-scale environments.

When managing large-scale networks, the time and the resources invested in the deployment and management of a solution can quickly outpace its benefits. If the deployment of a performance monitoring solution requires too many resources, then there’s little chance that it will ever be adopted. One technology that is facilitating deployment and maintenance are software containers. Containers are also getting supported by more hardware and software vendors. When evaluating a network performance monitoring solution for large-scale environments, make sure that the monitoring agent is available as a software container. It will drastically reduce the time to setup, maintenance, and also ensure that the benefits of the solution will exponentially exceed investments.

Author's Bio

Stefano Gridelli

CEO, NetBeez