The Concurrency settings page shows how execution slots are allocated to your organization by the platform’s fair scheduler. This helps you understand your organization’s processing capacity and how it compares to other tenants on the platform.
How Fair Scheduling Works
When multiple organizations are processing documents simultaneously, the platform uses a weighted fair scheduler to distribute execution slots equitably. Instead of a simple first-come-first-served queue (where one organization’s large batch could starve others), the scheduler considers each organization’s current load relative to its configured weight.
The core fairness metric is the load ratio:
load ratio = current in-flight slices / weight
The organization with the lowest load ratio gets the next available execution slot. This ensures that:
- Organizations with less active work get served quickly, even when another organization has thousands of queued items
- Organizations with higher weights receive proportionally more throughput
- No organization is completely starved, regardless of queue sizes
Example
| Organization | In-Flight | Weight | Load Ratio | Priority |
|---|
| Acme Corp | 2 | 1 | 2.0 | 3rd |
| Beta Inc | 0 | 1 | 0.0 | 1st (lowest) |
| Gamma LLC | 5 | 5 | 1.0 | 2nd |
In this scenario, Beta Inc gets the next slot because its ratio is 0. Despite Gamma having 5 in-flight slices, its high weight of 5 keeps its ratio at 1.0 — lower than Acme’s 2.0.
Viewing Your Concurrency Settings
Navigate to your organization’s Configuration tab and select the Concurrency card. You will see:
- Current In-Flight — The number of execution slices actively running for your organization right now. This value updates in real time.
- Load Ratio — Your current in-flight count divided by your weight. A lower ratio means higher priority for the next dispatch slot.
- Weight — Your organization’s relative scheduling weight. Higher values give proportionally more execution throughput.
- Max Concurrency — The hard cap on how many slices can run simultaneously for your organization. “No limit” means there is no cap beyond the platform’s global capacity.
Concurrency data appears after your organization has run at least one execution with the fair scheduler enabled. If you see “No concurrency data available”, contact your platform administrator.
Configuration
The Weight and Max Concurrency values are configured by platform administrators. If you need changes to your organization’s concurrency settings, contact your platform admin.
Weight
Weight controls your organization’s share of execution throughput relative to other organizations. The default weight is 1, meaning equal share.
| Weight | Effect |
|---|
| 1 | Standard share (default) |
| 2 | Double the throughput share |
| 5 | Five times the throughput share |
Weights are relative — an organization with weight 2 gets roughly twice the slots of an organization with weight 1, assuming both have queued work.
Max Concurrency
Max concurrency is a hard cap on how many execution slices can run simultaneously for your organization. This is useful for:
- Cost control — Limit parallel processing to manage compute costs
- Resource protection — Prevent one organization from consuming excessive platform resources
- Rate limiting — Match concurrency to downstream system capacity (e.g., API rate limits)
A value of 0 means no cap — your organization can use as many slots as the scheduler allocates based on weight.
Concurrency API
Organization concurrency settings can be read via the API:
| Operation | Endpoint | Access |
|---|
| Get org concurrency | GET /api/fair-scheduler/orgs/{orgId} | Authenticated users |
| List all orgs | GET /api/fair-scheduler/orgs | Platform admin only |
| Update org config | PUT /api/fair-scheduler/orgs/{orgId} | Platform admin only |
{
"organizationId": "abc-123-def-456",
"currentInFlight": 3,
"maxConcurrency": 0,
"weight": 1,
"ratio": 3.0
}
| Field | Description |
|---|
organizationId | The organization’s unique identifier |
currentInFlight | Number of execution slices currently running |
maxConcurrency | Hard cap on concurrent slices (0 = no limit) |
weight | Relative scheduling weight |
ratio | Current load ratio (currentInFlight / weight) |
Update Request
{
"weight": 2,
"maxConcurrency": 50
}
Only platform administrators can modify concurrency settings. The PUT endpoint requires platform admin privileges.
If your organization’s processing seems slower than expected during peak times, check your load ratio on the Concurrency page. A consistently high ratio compared to your weight may indicate you need a higher weight allocation — contact your platform administrator.