byasardell01-20-201708:32 AM - edited 01-24-201703:46 PM
Network automation encompasses a number of layers, including the control, management, and orchestration planes (Figure 1).
Figure 1: A Data Center Stack
Loosely, you can think of the layers as a vertical stack in this way, but there is nothing (but convention) to stop software advertised at one layer to perform the functionality of a different layer. And at times this will be desirable, to simplify a deployment and make it more robust.
Let’s assign broad definitions to the functions of the control, management, and orchestration planes:
At the control plane layer, SDN provides agility and automation, controlling groups of routers and switches
The management plane layer offers a wide array of tools, both traditional as well as DevOps style, for provisioning, troubleshooting, and remediating services across domains (monitoring and analytics processing frequently occurs here)
The orchestration plane exists above these layers to provide lifecycle management, capacity planning and a means to provision resources and define consumable services
SDN, as a design paradigm, can offer real-time programmability of network configuration and control plane behavior. Typically, SDN controllers are not focused on troubleshooting and remediation, and they don’t address management across compute and storage domains.
In general, orchestration products like OpenStack work across IT domains but do not address collaboration, monitoring, troubleshooting and remediation of the physical IT infrastructure–all areas that typically consume a large percentage of an engineer’s time.
DevOps-style automation tools can fill these gaps with a vast array of mix-and-match tools to provide collaboration and automation. These tools often complement (not replace) SDN and Orchestration deployments. Essentially, DevOps, from the management plane, can expand in both the control and orchestration directions.
The layers map to the software categories I’ve mentioned above. As organizations embark on their journey to digital transformation, they must investigate (through visibility) and act (through automation) through all layers.
Scanning from bottom to top in Figure 2, you can see the need for visibility from the hardware (where individual devices are controlled) to the orchestration layer (where cloud services are built):
In the networking hardware layer, we need pervasive visibility, perhaps facilitated through built-in architectures (such asSLX Insight or the Visibility Services), and we can also benefit from flexible, open platforms.
In the middle (workflows) layer, workflows need visibility from the network hardware layer, and must be event driven and customizable to maximize agility.
At the top (orchestration) layer, automation and visibility may be implemented in the form of plug-ins to relevant technologies such as OpenStack and other cloud management platforms.
In his blog on Using Visibility and Automation Together, Lindsay Hill illustrated how to use visibility and automation tools at the networking and workflow layers. In future postings, we’ll show how application orchestration is added to the mix for a complete picture of multiple automation layers.
Across IT Domains, from Technology to Culture
Now consider the horizontal plane in Figure 2. This is where it becomes apparent that a DevOps solution can be the lynchpin for the entire software stack. Related and complementary IT domains include the teams that manage applications, security, and storage.
And any successful digital transformation must include people, processes and policy.That's why another key aspect of managing your workflows is to have them work across these domains, and to accommodate the changing needs of the business.
Call to Action
For more information on how to use these technologies to digitally transform your business, follow the links in this blog, or contact your Brocade sales representative.