Esri Performance Benchmarks

Esri Professional Services conducts a series of performance validation tests in support of each Esri core software release. In addition, it evaluates Esri and third party products. These results are used to validate performance and scalability of Esri software and provide relative information for capacity planning for Esri based solutions. This section provides an overview of Esri’s testing process and an explanation of the benchmarks that can be found in the Implementation Gallery.

Esri Benchmarks

Esri Professional Services conducts and publishes Performance Benchmarks of representative test results in the Implementation Gallery. These benchmarks can help with:

System sizing
Capacity planning model calibration and input
Selection of optimal technology architectures
Selection of optimal application architecture

Without relevant performance benchmarks, it might be challenging or impossible to conduct effective capacity planning. In many cases, estimating with no test data, or using outdated capacity models can lead to a significant margin of error. This can result in degraded performance increasing hardware and license costs. An overview of the information found in the Performance Benchmark Reports can be found in the Capacity Planning and Performance Benchmarks Reference Guide.

Performance Benchmarks provide key information for sizing estimates and as an input for capacity planning, mainly:

CPU service time
Network Mbits/transaction
Machine SpecRate (for extrapolation of results for different hardware)
Max Throughput
Response Time for a give user load

Required Adjustments

The following key factors may need to be adjusted when applying Esri benchmark results.

Adjusting for workflow relative difference

Determining all details of the workflow during early planning phase might be difficult. However, for early capacity planning purpose, all operation information may not be needed, but focus on:

Transactions performed frequently with a relatively high response time
Batch operations - transactions with long response time

Adjusting for transaction “size” relative difference

Transaction “size” difference typically is related to the different:

Data source
For map service, ArcMap document content and configuration (number of layers, scale, symbology
Image compression
Map display size
Cached vs. dynamic services

Adjusting for Architecture difference

Many of Esri's benchmarks are conducted on a non-distributed, or Workgroup, test environment. Performance and scalability of this solution will differ from a distributed environment. For example, it may not be possible to extract the resource utilization of a specific component, e.g. SOC or Database from a benchmark conducted on a single server and apply to a distributed architecture. In addition, there may be a difference in the total number of CPU cores used which will impact the overall scalability of the system as measured by maximum throughput (TPH) .

In the case of a different workflow, capacity may be modeled using published CPU service times for individual transaction, if available.

Adjusting for relative CPU speed difference

This difference can be measure as a ratio of SpecRate CINT 2006 Base per Core published benchmarks. It should be noted that on the system running under relatively small load, the CPU speed, not the number of cores, will largely impact the performance of the system.

For more information, see Capacity Planning and Performance Benchmark Guide.

Addtional Support

Esri Professional Services offers a variety of services in these technical areas. Please contact Esri Professional Services.