PyU4V Usage Recommendations

In this section we aim to highlight various topics that we hope will improve your overall experience when using PyU4V. If you have any recommendations that you would like to make please let us know by opening an issue and we will review it for addition here.

Volume Naming Conventions

It is strongly recommended that you create a volume with a unique volume_name or volume_identifier. When you search for a volume device_id based on its volume_name, it is preferable to receive a single device id rather than a list of device ids, of which any could be the device that you just created.

Enhanced Performance Monitoring

If you are new to PyU4V it’s recommended rather than use the extensive list of functions in performance.py to use the Enhanced Performance monitoring functions in performance_enhanced.py. The enhanced performance functions provide the latest 5 minute diagnostic Key Performance Metrics for all components for the specified categories. These functions are simpler to use executing GET calls and don’t require input from the user for timestamps, t hey also provide bulk information on all components removing the need to perform lookup calls and execute loops to execute calls for each component object to retrieve metrics.

Enhanced Array Functions Bulk Data Retrieval

With PyU4V 10.0 and higher new functions for gathering filtering and selecting data for multiple objects such as storage groups and volumes. Using these calls reduces the number of API calls you need to get information on a number of objects and the amount of code needed to gather information on storage resources.

Performance Monitoring

When using PyU4V for performance metrics collection there are a number of best practices that you should follow:

  • After enabling Unisphere for performance metrics collection allow Unisphere 30 minutes to gather enough data before making any calls.

  • The most granular time available with PyU4V performance metrics collection is 5 minutes, so querying for data more frequently than 5 minutes is a wasteful use of resources.

  • If you want to ensure that your performance metric collection is querying the most recent performance data available set a recency window value on your calls.

  • If the performance timestamp is not recent as of 5-10 minutes ago there is a strong likelihood that your instance of Unisphere has gone into catchup mode and is processing a backlog of performance data. It will resume normal operations once this backlog processing is complete and be indicated by performance timestamps with a window of 5-10 minutes from the current time.

  • When querying a single instance of Unisphere for performance metrics across a number of arrays be careful on the load placed on Unisphere and try to determine if it is possible to query that amount of data in the given time frame you have set.

  • If querying for data at regular intervals, examine your calls to see if you can create the interval time. If information is not likely to change over the course of 24 hours then querying once a day would be sufficient.

  • If querying for real-time performance data it not possible to query for less than one minute of data at a time, this is to try negate the possibility of users querying for real-time data every 5 seconds and overloading Unisphere. With one minute intervals users can query for 12 deltas of data at 5 second intervals for a specified minute range.

  • It is not possible to query for more than one hour of real-time performance data at a time. This is a hard restriction enforced by Unisphere. If you need more than one hour of data then consider if diagnostic performance data is more suitable.

Lastly, and most importantly, with great power comes great responsibility, PyU4V provides you with the ability to query every performance metric for every performance category. Instead of gathering everything possible, be resourceful with your calls and only query what is needed. This will provide improvements in PyU4V performance, network load, and Unisphere REST performance. If you are only interested in querying for KPIs, you can specify that only KPI metrics are returned, but better still only query for a subset of metrics that you are interested in.