Enhancing infrastructure management capabilities

Analyzing near-term capacity trends

• Assign more volumes manually.
• Create resource assignment rules based on your evaluation of the risk report data.
• Develop a plan for redistributing storage resources by moving storage pools to different volumes.

Long-term capacity planning

Once you get an idea for the trend pattern for the various consumers or application users, create a risk profile that specifies capacity metrics. To maximize the information shown in the trend projection, create one profile for one month, one for three months, and one for six months. Use these profiles in conjunction with report definition that includes consumers.

Near-term performance tuning

You can use predictive analytics risk reporting to fine-tune overall system performance over time to achieve optimization. Because most large-scale IT infrastructures are heterogenous, the task of fine-tuning performance is ongoing. Infrastructure administrators might have the twin goals to make the most of the existing IT equipment (all system resources), while managing resources to achieve stability in performance. With the predictive analytics risk reporting feature in Ops Center Analyzer, you can generate reports with trend projections to anticipate performance fluctuations and adjust resource monitoring.

In this workflow, infrastructure administrators can fine-tune system resource performance and user resource allocation on a weekly or monthly basis.

1. Set up dynamic thresholds for user resources. Use the base dynamic threshold profile or edit it for the following:
   • Metrics: If you want to track performance at a granular level, select more metrics.
   • Plan: You can adjust the profile to monitor during peak times during the week.
2. Set up event notifications.
3. Create risk profiles and report definitions to the corresponding metrics and resources in the profiles.
4. Generate a risk report along with standard reports to compare past and current trends with the trend projections in the risk report.
5. Make adjustments in resource allocation, and modify both threshold profiles as needed.
6. Generate the next round of risk reports and determine if the adjustments averted overutilization of resources.

As an ongoing process, infrastructure administrators have several options on how to act on the information in these risk reports. Ops Center Analyzer offers the following tools to aid in fine-tuning:

• You can develop a script that is invoked when certain performance events and threshold violations occur, and use the Execute Action function to run that script.
• You can use Execute Action to invoke a service from Ops Center Automator.
• If the magnitude of your infrastructure requires thorough analysis of data to avoid service interruptions, you can do the following:
  • To analyze minute-level data, adjust the collection time in Analyzer detail view to track events in finer granularity.
  • To analyze second-level data in Hitachi resources from Ops Center Analyzer, run granular data collection.

Damage Control

You can use the predictive analytics risk reporting feature to evaluate near-term trends and take preventive measures against performance degradation in your infrastructure.

Performance degradation can affect infrastructure with sudden bottlenecks. or worse, extended outages. Consumers might experience I/O problems with lagging response times, an annoying occurrence to application users, and a headache for IT. However, in an extended outage, the application is no longer available for use, causing a work stoppage. This situation can be more than a headache to IT and might result in escalating support calls.

Use risk reporting as an extra layer of monitoring

You can add an extra layer of resource monitoring to give a 10% margin outside your normal threshold limits during day-do-day operations by using predictive analytics risk reporting. To manage performance risks on a daily level, you can generate risk reports to establish a buffer for certain thresholds. Doing so allows infrastructure administrators to predict when thresholds violations will occur. This buffer makes it easier to react to performance problems.