Loading

Understand Salesforce Org Throttling

Date de publication: May 16, 2025
Description

To manage capacity and availability, Salesforce sometimes applies a throttle to your org in Sandbox or Production instances. Requests are placed in a queue and executed at a slower rate compared to the incoming request rate, typically 50% of the incoming rate of requests. The impact on your org depends on your implementation and can include increased request times, latency, or timeout errors.

Salesforce has automation and recurring processes that increase horizontal and vertical scale. Throttles are applied when these processes are exhausted and guardrails aren’t sufficient. Customer success is a shared responsibility and Salesforce regularly notifies customers about phases in the customer lifecycle journey. 

Résolution

Common Reasons for an Org Throttle

  • Execution of an inefficient SOQL
  • Unapproved performance testing, as performance testing is only allowed on Sandbox instances via the performance testing approval process
  • A sudden spike in inbound synchronous requests per second that includes browser requests and SOAP or REST API calls
  • A high number of inbound asynchronous requests per second that include Bulk API jobs, @future jobs, and Apex Queueable jobs
  • Rare Salesforce infrastructure issues that cause automatic or manual throttles


How a Throttle Works
Salesforce is a multi-tenant cloud platform and ensuring all customers’ success, scale, and performance is a mutual responsibility for customers and Salesforce. Salesforce’s automatic throttle algorithm looks for the top resource consumers by org and feature, and attempts to be as precise and accurate as possible. The top resource-using customer is most often the cause of high resource utilization. 

We’re actively working on improving this process to ensure the least amount of impact on the customer experience of the throttled org. Salesforce is also using AI to predict resource overutilization and apply AI to dynamically adjust throttle rates and maximize the customer experience.

For all customers, an outbound support case is created that contains the following details about the throttle.

  • When the throttle was applied
  • Features that were throttled such as URL, aura, and services
  • Throttle rate
  • Resource utilization such as DB CPU and DB Conn Pool (Signature customers only)
  • Request rate and number of requests (Signature customers only)
  • Throttle causation analysis (Signature customers only)
Preventing an Org Throttle 
  1. Improve the SOQL performance by making the query selective. 
    1. Follow the guidance in the Well-Architected framework or work with a Success Architect. See Performance Patterns and Anti-Patterns. 
    2. See the Knowledge Article Make SOQL query selective.
  2. To identify inefficiencies and bottlenecks, do performance and scale testing in Sandbox or PTPaaS. Get Sandbox instance performance testing approval. 
    1. See Performance Testing FAQ.
    2. See Salesforce Help: Scale Center.
  3. Investigate large numbers of inbound requests and use scaling features like waiting rooms.
    1. For Signature customers, if events like flash sales or open enrollment are causing large numbers of requests, consider working with the Customer Success Manager (CSM) to enroll in the Key Event Management process.
    2. If the requests are related to a Denial of Service, set up DOS mitigation.
  4. Make asynchronous requests or batch them when possible, such as Bulk API.
  5. Set up monitoring and alerting in Production and prepare incident response plans for when the org gets throttled. Consider Signature Success or Premier Support for enhanced support and features. See Proactive Alert Monitoring: Database CPU Consumption Time.
  6. If none of the above apply, contact customer support. Internal Salesforce infrastructure issues are posted to trust.salesforce.com. 
Numéro d’article de la base de connaissances

000384939

 
Chargement
Salesforce Help | Article