PBS Analytics Features

PBS Analytics allows users of HPC systems to gain insights into how well expensive HPC resources are being used, to inform better planning and purchase decision-making.

  • NEW in 13.0: Chart and dashboards sharing — within new “Shared” category, users with special privileges can create, modify, and delete charts and dashboards (all other users have viewing access).
  • NEW in 13.0: Optional node configuration improves installation process — with PBS Analytics 13.0, node configuration is no longer necessary — so installation is quick and easy. System administrators can install and then supply node information whenever they desire.
  • NEW in 13.0: New parsing and caching progress indicator screen — the Web Application has a new Progress Indicator Screen to visualize the progression of the parsing and caching process.
  • More flexibility in creating clean, concise charts, including additional chart options (such as XYZ plots), to furnish a deeper understanding of organization’s HPC infrastructure and better root-cause analysis with powerful chart designer
  • Ability to track activity by job, software, hardware, user, project, node class, etc.
  • Ability to track “custom” resources (i.e. node classes, GPUs, licenses, scratch space, etc.)
  • Ability to track hardware utilization, including accounting for configuration changes over time (i.e. adding/deleting nodes; downtime)
  • Insights into user efficiency and productivity based on usage of resources, job success rates, etc. — so user efficiency issues can be identified and corrected
  • Insights into group productivity, so users can be added/moved to/from different groups as necessary to optimize productivity
  • Ability to display statistics within an operations room or headquarters reception with PBS Analytics’ slideshow capability
  • Ability to visualize resource Demand and Supply (e.g. Available cores, Queued Cores and Used Cored)
  • Views into totals, counts, times, averages, etc.
  • Support for PBS clusters in different time zones (accounting logs from various PBS clusters located in different time zones can be parsed and PBSA will display all dates in a uniform time zone)

Depth and Robustness
  • NEW in 13.0: Stability and scalability — database size is reduced, so PBS Analytics 13.0 scales nicely with more cores and memory for bigger datasets. An optimized HPC domain-aware query engine improves query performance and database resource usage. Also, the parser now runs as a separate, leaner component – it is easier to restart the Parser without affecting other components.
  • NEW in 13.0: Revamped custom resource hook — the custom resource configuration file has been converted to an XML format, making cus¬tom resource definition less error prone.
  • Improved logging: PBS Analytics 13.0 offers improved logging of error messages, including the reasons for discarding a job while parsing the PBS Professional accounting logs.
  • NEW in 13.0: JDK update – PBS Analytics now uses Java 8, enabling improved security.
  • NEW in 13.0: Database schema changes improve parsing and caching times — changes to the PBSA database schema have improved the time required to parse and cache the PBS Professional accounting log data.
  • NEW in 13.0: Documentation has been improved based on various field and customer inputs, and we have created new videos for major features. Videos can be found in documentation folder at the GA location.
  • Numerous chart types, including pie, line, bar, 3D, radar, scatter, etc.
  • Drill-down to underlying data and export to Excel
  • Ability to export Chart and Dashboard as Image or PDF
  • Ability to draw percentage (%) of resource utilization
  • Support for billing for node wall time
  • Support for PBS peering configuration — peer scheduling allows separate PBS complexes to automatically run jobs from each other’s queues; this means PBS can dynamically balance the workload across multiple as well as separate complexes. PBS Analytics now reports peer-scheduled job information.
  • Secure Socket Layer Support: Secure transmission of accounting logs from Data Collectors to the Analytics Engine; the “https” protocol is configured by default
  • Backward compatibility – Reports that are saved using older versions of PBSA can be retrieved through migration tool kit
Simplicity and Ease of Use
  • NEW in 13.0: Reduced installation and configuration time — with PBS Analytics 13.0, numerous improvements make installation quicker and reduce time needed for post-installation configurations.
  • NEW in 13.0: Manual configuration eliminated — thanks to improvements in job data export, the PBS Analytics 13.0 Web Application displays all relevant job data used to build a chart when a user clicks the Job Data button, or exports the job data to a comma separated values file.
  • NEW in 13.0: Ease of maintenance — users can reset PBS Analytics 13.0 including all Data Collectors — the pbsa-data-reset command now automatically resets all Data Collectors making the process for discarding parsed accounting log data straightforward and uncomplicated; there is no need to perform a data reset when adding / modifying / tuning node details and holiday files.
  • Ability to visualize multiple different charts on one display with dashboards
  • Fast, easy analysis of large quantities of usage data with a high-performance database and robust data collectors
  • Ability to consolidate data by top (or bottom) usage, e.g., “Top 10”
  • Ability to filter by user, group, project, queue, host, etc.
  • Quick access to canned charts out-of-the-box – and it’s easy to customize your own
  • Aggregation of data from multiple PBS Professional servers
  • Ability to customize, save, and share charts or entire dashboards

Request Information

Get Started Today!