Automating Network Operations for Greater Agility

Editor’s Note: Occasionally, Enterprise Networking Planet is proud to run guest posts from authors in the field. Today, Juniper Networks’ Harsh Singh, a networking industry veteran with over 15 years of experience in the field, walks us through what’s necessary to create an agile network through automation and orchestration.

By Harsh Singh

Today’s business world demands greater agility than ever before, forcing a transformation in enterprise data centers. Ever heard of virtualization? The cloud? These sweeping evolutions demand a corresponding evolution in data center infrastructure. Businesses that fail to keep up risk losing customers and revenue to their more agile competition.

Over the last decade, server virtualization has changed the way businesses build and manage their data centers. Server virtualization offered the immediate benefits of improved resource utilization, resulting in a reduction in operational expenses. And over time, businesses began to realize a secondary benefit that far surpassed mere opex reduction: agility.

Today, enterprises can spin up servers in a matter of minutes, instead of waiting the weeks or months the process used to require. Unfortunately, the network provisioning process has not kept pace. Simple configuration changes can still take days or weeks, mainly because provisioning and configuring the network remain primarily manual. Consequently, the network has become the bottleneck, preventing companies from achieving not only a highly responsive data center infrastructure, but also an agile business.

Data center operations basically comprise three tasks:

  • Provisioning, the initial deployment of networking gear;
  • Management, the day-to-day administration of the network, which includes tasks like the changing of passwords; and
  • Orchestration, the management of different elements of the data center infrastructure in concert with each other.

Automating these tasks results in a data center that can adapt and rapidly respond to a changing business environment. Let’s look closely at each of these tasks to see how we can begin to automate them.

Automating network provisioning through zero-touch provisioning

Network provisioning can be complicated. Typically, a top-of-rack switch deployed in the data center is first staged, a process in which the correct image and configuration are loaded, before being placed in production. Configuring the network of a data center with 1,000 servers requires hundreds of thousands of lines of CLI. As a result, provisioning the network takes not only time, but the work of highly trained professionals, making the process not only lengthy, but also expensive and prone to human error. A feature called zero-touch provisioning (ZTP) can automate this process.

ZTP allows you to load the image and configuration of a switch in a central location, typically a DHCP or TFTP server. A switch is then racked, stacked, and powered on. When powering up, the switch sends out a DHCP request and receives the location of its images and configuration in response. The switch then downloads the image and configuration and becomes fully operational. This process is repeatable and customizable to different switches. ZTP can dramatically reduce the time to provision the network in a data center and decrease network downtime caused by human error.

Scripting and programmability to automate network management

Once provisioned, switches must be managed on a day-to-day basis. This work takes up the bulk of a network administrator’s time, hence the need for automation.

Switches need a flexible scripting mechanism to allow them to automatically respond to simple alerts and events. This mechanism will ensure higher uptime and faster average time to repair, resulting in greater agility and improved customer experience.

In addition to scripts, switches also need open programmable APIs to allow external tools to programmatically configure the network. Traditionally, tools had to ssh the switch and do something called screen scraping, a process that read the output on the screen. This process is error-prone and requires retooling every time the CLIs change. By providing an open programmable API, external tools can now automatically configure the switch and provide better workflow integration.

Achieving data center orchestration

Data centers are an ecosystem of compute, storage, and networking equipment. Unfortunately, today most organizations split their data center management up into separate entities, each run by a separate team. This has resulted in a rigid infrastructure that struggles to adapt to an ever-changing business environment. Virtual machines can now be spun on demand, but the requisite VLAN provisioning on the network takes days because of an archaic change request process. Creating a truly agile data center infrastructure requires management of different elements of the data center in concert with each other and across silos of teams.

Changes to network configuration need to happen automatically and immediately. For example, when a new virtual machine is spun up, the VLANs in which it belongs should have automatic configuration on the top-of-rack switch. To achieve this kind of integration, switches must support orchestration tools like OpenStack and CloudStack. In addition, support for IT automation solutions like Puppet or Chef will enable network management in tandem with other data center infrastructure.

Data center automation and orchestration form the lynchpins of the agility of any business. When considering networking solutions for the data center, look for those designed with a flexible and open standards-based framework. Additionally, seek out solutions that simplify the data center by enabling automation across the full operations lifecycle, from network provisioning to management and orchestration. This will streamline processes and reduce the risk of human error. A networking solution that does these things will allow an enterprise to capitalize on the benefits network automation has to offer today and will enable organizations to implement the new and emerging technologies of the future, like software defined networking (SDN), as dictated by evolving business needs.

Harsh Singh of Juniper NetworksHarsh Singh is Senior Product Marketing Manager of Data Center and Cloud Infrastructure at Juniper Networks.

Latest Articles

Follow Us On Social Media

Explore More