Network Programmability, Provisioning, Automation & Orchestration

Effective data centers are central to the success of any enterprise, and the flexibility and agility of data center operations is essential to successful mission and business outcomes. As technology has evolved, so too have data center capabilities, including the introduction of virtualization strategies that have enabled some level of agility and flexibility. However, network operations still tend to rely on many manual processes that require the expertise of highly-trained professionals and may be subject to vulnerabilities introduced by human error. These operational realities can lead to higher data center management costs and, in some cases, loss of business or missed opportunities due to lack of agility or data center downtime.

Consequently, An Enterprise operations strategy should be based on automating provisioning, streamlining day-to-day management, and orchestration of networking platforms that combine to enable faster response to dynamic business environments. In addition, New switching platforms and management tools provide open interfaces to integrate seamlessly with third-party automation and orchestration tools, so that network operations are integral to the data center workflow.

Data Center Evolutions and Trends

As a foundational enterprise component, data centers are continuously evolving to meet the demands of business, mission, and operational requirements. Data centers have undergone a series of transformations, from the days of mainframe computing to the client-server PC era to a world of distributed applications. Today, applications are defining how data centers are designed and operated. Applications management must provide capabilities for applications to be quickly brought up, brought down, and moved to respond to changing conditions and to make best use of underlying resources. To meet the growing need for agility and flexibility, the data center components (compute, network and storage) also need to evolve and provide a level of automation and orchestration that can meet present and future demands.

Need for Network Automation

Server virtualization initially was designed to improve resource utilization. It has quickly evolved to provide a server environment that is more agile, a significant benefit that is proving to be increasingly critical in the efficient data center. In the past, deploying a server could take days, if not weeks. Today, virtual machines and applications can be deployed in seconds, helping organizations meet their business requirements while reducing operational expenses.

This agility comes from the fact that the compute infrastructure has been integrated with external automation tools and management systems. However, a data center is an ecosystem of multiple elements, and changes made to the compute infrastructure require corresponding updates to the network infrastructure. Unfortunately, the provisioning and management of the network often has not kept pace with those of virtual machines and applications. It can still take days, if not weeks, to provision the required flow across an entire network, mainly due to the fact that network provisioning and configuration are primarily manual processes. For many organizations, the network has become a bottleneck, preventing those enterprises from deploying a robust, agile, and highly responsive data center infrastructure.

Network Operations and Automation

Data center network operations can be categorized as three specific tasks: provisioning, management, and orchestration. Provisioning involves the initial deployment of networking equipment; once provisioning is complete, management covers the day-to-day operational decisions, and lastly, orchestration focuses on the ability to manage the different data center elements, servers and storage in concert with each other. To implement a data center that can adapt and respond rapidly to changing business and environmental conditions, it is necessary to automate all these tasks. Let’s look more closely at each of these tasks and understand how we can begin to automate them.

Provisioning: Network provisioning is a complicated process. Typically, a top-of-rack (ToR) switch that is deployed in the data center is staged, a process where the correct image and configuration is loaded before it is placed into production. Usually, this is a manual process, performed by trained networking professionals. As a result, provisioning the network is not only time-consuming, but also error-prone. This initial process can clearly benefit from automation where switches are able to download the correct image and configuration as soon as they are racked, stacked and powered on.

Management: Once a switch is provisioned, it needs to be constantly managed. This is where the bulk of a network administrator’s time is spent and where automation can make a real difference. The switches need to have a flexible scripting mechanism that allows them to respond automatically to simple alerts and events. In addition to scripts, switches also need to be programmable by third-party tools which will allow for better integration into the data center workflow.

Orchestration: Data centers are an ecosystem of compute, storage, and networking equipment. Unfortunately, these elements often are managed by different teams as separate entities. This approach has resulted in a rigid infrastructure that is not agile enough to adapt to an ever-changing business environment. As an example, today Virtual Machines can be spun up on demand, but the required VLAN provisioning on the network may take several days because of an archaic change request process. Creating a truly agile data center infrastructure requires that different elements of the data center are managed in concert and across these subject-matter teams.

Ideally, network configuration changes need to happen automatically and immediately. When a new Virtual Machine is spun up, the VLANs it belongs in need to be configured automatically on the ToR switch. To achieve this kind of integration, it is imperative that switches provide support for orchestration tools such as OpenStack and CloudStack. In addition, by supporting IT automation solutions like Puppet or Chef, the network now can be managed in concert with the overall data center infrastructure.