Datacenters remain critical for Enterprises and Service Providers, even in the era of hybrid and multi-cloud. Data centers are not only critical internal resources for a company, in a period of digital transformation, they are often critical customer facing resources too.
Operations teams face overwhelming complexity and data volume. They cannot process in a timely manner all the information provided to them. Therefore, teams can no longer be reactive to every incident; they must now reduce the number of incidents, by being proactive. Other challenges include multilayer complexity and multivendor management.
The datacenter operations challenge is different than managing a public cloud software deployment. Operations teams must manage physical and virtual infrastructure, including network overlays and underlays.
For decades, the mantra has been “if you can’t measure it, you can’t manage it”. Today, operations teams can measure much more than they once could. The sentiment is now shifting: “Just because you can measure it, does not mean you can see it”. Both the number of network endpoints and the volume of data per endpoint is increasing. Operations teams are today challenged by significant and growing noise.
Clarity about present network conditions is critical. Equally critical is clarity about future network conditions. As operations teams orient themselves toward proactive approaches, the less reactive they will need to be, reducing the impact on applications and network services from changing network conditions.
Agility is desirable at all layers and phases of deployment. If one equipment or software supplier is the best option for a specific network domain, or even a new POD, then operations must be able to support multivendor approaches. Independent operations tools vendors have an inherent interest and motivation to develop and support multivendor solutions.
Dense datacenter topologies have created significant challenges for operations teams, and spurred industry innovation in routing protocols. Some underlays are based on a traditional interior gateway protocol (IGP) like OSPF, while BGP4 has emerged as a popular underlay with the adoption of Clos fabrics in the data center. BGP4 has also emerged as a popular overlay due to the increasing adoption of EVPN. In addition, physical layer equipment, links, and optics require management. Effective operations tools must provide an integrated and coherent approach to multilayer operation.
New dense topologies require operations tools that integrate topology. When a problem arises, operations tools must understand where in the topology the problem is occurring, and what the consequences are. Many operations tools simply collect and display data. Some go beyond that. There are a limited number of solutions that were designed from the outset to integrate topology into the model.
Data centers are critical to Enterprises and Service Providers. However, the benefits of dense, fixed-form factor architectures and increased availability of data have created new challenges. Operations tools must respond to these challenges by reducing information to operationally relevant, actionable, proactive insights across multiple layers and multiple vendors.