Frequently Asked Questions (FAQs) about PI System topologies

Q: Why have the PI System topologies and performance envelopes changed?

A: The most recent topology updates are due to testing hardware upgrades in the test lab environment. AVEVA continues to make significant investments to the software and design of the PI System. Simultaneously, hardware and virtualization technology continue to advance in speed, scope, and capabilities. The topologies are kept up to date to reflect these changes. 

Q:  The performance envelope numbers seem low for the hardware assigned.

A:  The topology performance envelope numbers do not reflect the maximum performance capabilities of the PI System.  The performance envelope numbers are intended for planning and comparison with the expected steady state operation of the production system. The goal is to ensure reliability and quiet enjoyment of the PI System, while being mindful of infrastructure costs.  

Headroom has been left to handle upset conditions or unusual interest in the stored data. Actual maximum capabilities are above the published numbers for individual components, and are virtually unlimited in a federated PI System that is managed through PI Asset Framework (AF). 

Q. What hardware was used in the test lab environment?

A. The topologies are tested on a virtualized environment. The following table lists the specifications for all topology deployment environments: 

Q. What impacts do disk size and speed have on the performance of a topology?

A. Disk size has minimal impact on topology performance and can be freely adjusted to achieve the required data archive storage (see Determine the disk volume requirements). In testing, topology PI archive hard disks have been benchmarked as follows: 

• Random 4KB Read, Queue Depth 16: 226.26 MB/s

• Read Latency: 0.28 ms

• Random 4k Write, Queue Depth 16: 112.44 MB/s

• Write Latency: 0.56 ms

  Note: Prioritize disks with sub-millisecond latency for the PI archive drive since this is more effective than disk throughput and input/output operations per second (IOPS). For example, Microsoft Azure Premium SSD v2 disks were found suitable for Data Archive workloads.

Q.  What happens if I get a burst of notifications that exceed the published notification generation rate for the topology?  

A. PI Notifications is capable of handling short bursts of notifications at much higher rates than is published in the performance envelope. The performance envelope rate provides a baseline for typical, routine alert scenarios. Users can generally expect notifications to be sent in under 5 seconds for most of these typical scenarios. During a burst of notifications, alert time may be increased. 

Note: Notifications are triggered by event frames. As a best practice, follow the guidelines set for event frames generation when designing and implementing your notification and alert strategy.

Q. What are the implications of generating large numbers of event frames?

A. Event frames are stored in the PI Asset Framework (PIFD) database on Microsoft SQL Server. As with any large SQL database, performance ramifications can arise if the database does not have proper resources. For more information, please visit the Knowledge Base page on the AVEVA Support portal.

Contact AVEVA Customer Support or your local AVEVA customer representative if you have other questions.