Data Center Project Best Practices #1: Assessment by Dan Klimke
Network managers and engineers in "Best in Class" data centers follow a well-defined project plan to implement change so as to more effectively transition with no or limited downtime, and deliver on project objectives. At a high level, the "lifecycle" of these projects can be divided into these three stages...Assessment, Installation and Validation, and Troubleshooting and Maintenance. This is the first of three posts exploring each stage and some of the 'success factors' or essential requirements needed for each step. We'll discuss what tasks and information are required at each stage, and how different features of the OptiView XG can provide the critical data required at each point in the lifecycle.
Before any change is made, the data center should be assessed. During this stage, the current state of a service, application or system needs to be evaluated, providing a performance baseline that can be later used for comparison during the validation stage. In some environments, the excitement of installing a new service or upgrade has led some to skip the assessment step in the lifecycle. But as many have found, taking an assessment has saved them from headaches during the installation stage, including time and money that can be wasted simply because the current system is not understood.
This assessment should include evaluating the flexibility and scalability of the infrastructure and servers, identifying and researching changes that need to be made – whether in the software, hardware, or virtual environment – and finally, evaluating potential impacts of proposed changes. To meet these goals, it is important to take a complete inventory of the components that will be impacted by the change, including the network infrastructure and servers. An up-to-the-minute network topology should be collected, to ensure that the whole environment is correctly understood as-built. A performance baseline of key link utilization, server response time, and server resource load should be conducted as a starting point to measure the current system. Finally, the data center documentation should be updated pre-change so that accurate records are available - these may become critical in the event of an unexpected outage or service degradation during the change.
In the assessment stage, the OptiView XG can play a strong role in baselining and measuring network and server performance before changes are made. Using the Key Device testing feature, infrastructure devices and servers can be measured over time to find delays that may occur over the course of a full day, and shows changes in performance to be verified post-change. The OptiView also will automate the discovery process, finding every device, interface and path through the data center, making sure that documentation is kept up to the minute should any unexpected problems arise during installation. Also, before links are upgraded, the present bandwidth can be measured at speeds up to 10Gbps, at full line-rate. Additionally, monitoring traffic on key links can identify what users and applications are taking up the bandwidth, which may result in an adjustment in allowed traffic, or a restriction in the bandwidth provided for non-critical applications.
The saying that "the only thing constant is change" is particularly true in the data center. But by following a well-designed project plan, and having the right information needed at each stage of the project's lifecycle, changes can be implemented successfully with little or no downtime. The OptiView XG was designed to give network and application engineers in the data center the critical information they need to ASSESS, VALIDATE and TROUBLESHOOT when making changes in the data center.