Why Every Network Needs a Doctor
Every year, I go to my doctor for my annual physical. My doctor goes through a standard series of procedures every time. He asks me questions about my diet and my general physical well-being. He puts his cold stethoscope to my chest and listens to my heart and breathing. He checks my blood pressure and heart rate. His lab technician takes samples of blood for tests to collect information related to organs like my liver and pancreas. He may even request tests depending on what he has learned during the exam. Later, my doctor follows up to discuss the results. He may tell me that I am perfectly healthy, or that I need to take a medication for a period of time, or to make some dietary changes based on the results.
Through his education, learning, and experience, he is able to take inputs from different components that, on their own, seem to have no relation to each other. In reality, there are many interdependencies with the different parts of my body and the statistics that were collected during the exam. Every organ and function in my body impacts the functions of every other system.
My doctor is the management and orchestration system for my physical well-being. He periodically collects disparate pieces of information and is able to understand the relevance between them to determine my overall health. If necessary, he makes adjustments to correct my state to bring my body back to a normal situation. Like the human body, your network has many sophisticated interdependencies between the various functions. It too, requires a ‘doctor’ to manage and orchestrate its health and well-being.
Maintaining healthy applications
Today, we are focused on ensuring application service level assurance. The network infrastructure is a framework to enable applications in the environment. The consumers of the applications are concerned about the functional delivery of the application, not how it is delivered. Guaranteeing application SLAs requires a network-wide holistic approach treating the network architecture as an interconnected ecosystem of individual components.
Today’s networks are evolving to a virtualized model utilizing cloud technologies and incorporating new virtualization architectures like SDN and NFV. The virtualized network is designed to be agile and elastic. They offer the ability to adjust the availability and scalability of the application based on changing conditions. This is designed to provide a more agile and flexible framework to ensure application delivery.
This capability requires the monitoring of the multiple points in the network to determine when changes need to be made to the network topology and application infrastructure. Statistics need to be collected from the multiple points in the IT architecture. These statistics need to be heuristically analyzed and a business and application specific policy needs to be applied to determine the current state. If the state is outside the defined normal parameters, adjustments need to be made to correct the deviation.
The management and orchestration system needs analytics – the collection of data from different points in the network from different technologies. It needs heuristics – the intelligence to understand the relationships between these different metrics based on each network’s unique design. Finally, the system must have orchestration and automation – apply the information learned to custom user defined polices so it can enact mitigation actions to reconfigure the network for the changing conditions.
My doctor is my medical orchestrator
This is exactly like my doctor as he maintains my physical well-being. My body is a biological ecosystem. Personally, I do not care about all the details and the actual values of all the tests performed during my physical examination. The important point for me to understand is that my body is either healthy or that I may need to take some medication that my doctor prescribes to bring it back to health. This is because he has the appropriate data (analytics), knowledge to interpret that data (heuristics), and has delivered an appropriate response to me about my system (orchestration and automation).
As our networks continue to become more flexible and dynamic, we need a component to maintain a consistent delivery of application service levels that people expect. The management and orchestration system within the network is becoming the equivalent of my personal physician in the same way that our network architectures are becoming the equivalent of our biological models and ecosystems.
The virtualization of the network architecture and focus on application delivery makes the necessary fine tuning and continuous adjustments harder to manage manually. An overarching management and orchestration system based on the understanding of application service level assurance is essential.