Standard validation checkpoints are a time-consuming, yet instrumental, part of day zero FlexPod configuration. To build in validation checkpoints, configuration is broken down into groups of steps. After each set is complete, the implementation engineer performs cross validations to ensure the build is moving in the right direction. If the build passes the validation exercise, the implementation engineer can move onto the next step. If it doesn’t, the problem must be resolved before further configuration can take place. Building a FlexPod is like constructing a house: There’s a logical order, and a sound foundation and frame must be in place before the walls go up.
For example, virtual port channel (vPC) is a critical checkpoint. If the virtual port channels(vPC) are not configured correctly, difficulties will arise with network topology and end devices down the road. To check if a virtual port channel is deployed correctly, implementation engineers can create a vPC domain with peer-link configurations, create appropriate VLAN trunks, and then ensure the endpoint port channels are formed successfully. (For FlexPod configuration, endpoints are typically Cisco UCS network uplinks and NetApp FAS ethernet ports that require etherchanneling across two different Cisco Nexus Switches.) If the endpoint port channels are functional, then it’s safe to continue building the FlexPod.
Because each element that is configured builds on the last, checkpoints minimize the chances that at the end of the configuration process, the deployment engineer will realize he’s got an invalid setup . When that happens, the deployment engineer has to retrace his or her steps, mining through hundreds of configuration variables to find the error, and then repeat the work that has just been undone.
Defining and validating the configuration can be quite challenging, as well as time consuming. Configuration checkpoints add anywhere from several hours up to several days, depending on the complexity of the build and how much troubleshooting is required. When configuration checks fail, even more time is lost. The implementation engineer must refer back to the admin guides for troubleshooting, or escalate to an internal expert or to the OEM. In some cases, the VAR must send an expert who specializes in the problematic hardware to the customer site to assist the implementation engineer.
These are all scenarios that I encountered in my previous role as a solutions architect. They were frustrating, but I believed they were inherent to the FlexPod build process. I came to StrataCloud because I realized that these problems were not in fact inherent, and that they could be solved through software. I set out to help our engineers incorporate features into SDI Install that could solve some of the most frustrating problems I had encountered FlexPod build process, and manual validation topped the list. With SDI Install, we have managed to automate validation, and not only that — we’ve removed it entirely from the deployment engineer’s to-do list. Let me explain.
By automating the validation process, SDI Install eliminates the need to build time-consuming checkpoints into the configuration process. SDI Install validates the design in real time before configuration begins. If errors are found, SDI Install provides the deployment engineer with the appropriate action for remediation. That means that most of the time, the engineer can resolve potential configuration issues on his own before configuration begins, instead of having to go through a time-consuming troubleshooting process. This preventive measure both empowers the implementation engineer to complete the FlexPod build from start to finish and ensures that once configuration is launched, it runs smoothly and is completed on time.