WCUC

Computer & IT

Website Concurrent User Capacity Calculator

Estimate safe website concurrency from tested application throughput, CDN offload, per-user activity, database query intensity, and active-session capacity.

Safe application throughput (RPS)-
Origin requests per active user per second-
Concurrent users by application tier-
Database queries per active user per second-
Concurrent users by database tier-
Safe concurrent active users-
Capacity margin versus forecast-
Sustainable new sessions per minute-
Forecast application traffic (RPS)-
Forecast database traffic (queries/s)-

Decision view

Concurrent-user demand curve across application, database, and session ceilings

Concurrent-user demand curve across application, database, and session ceilingsA live demand line crosses three tier limits and marks the first bottleneck before forecast traffic.
Exact scenario comparisonRequests served without the application (%) changes while all other entered assumptions remain constant.
Requests served without the application (%)Safe application throughput (RPS)Origin requests per active user per secondConcurrent users by application tierDatabase queries per active user per secondConcurrent users by database tierSafe concurrent active usersCapacity margin versus forecastSustainable new sessions per minuteForecast application traffic (RPS)Forecast database traffic (queries/s)

How to use Website Concurrent User Capacity Calculator

  1. Measure requests per active user from real browser sessions.
  2. Use safe, not maximum, application and database throughput.
  3. Enter concurrent active users rather than logged-in accounts.

Calculator guide

Understanding Website Concurrent User Capacity Calculator

Requests per second and concurrent users are different quantities. This calculator converts one active user's request behavior into application and database load before comparing those tiers with the session-store limit.

Behavior converts units Per-user RPS connects throughput to concurrency.
Database fan-out Queries per origin request create a second load path.
State matters The session store can bind even when compute has room.

Detailed calculation process

Detailed website concurrent-user calculation

The default site has six instances rated at 280 RPS, uses a 70% target, offloads 35%, and models nine requests per active user per minute.

General formula: Q_a=N*q*uq_u=r/60*(1-c)C_a=floor(Q_a/q_u)q_db=q_u*dC_db=floor(Q_db/q_db)C=min(C_a,C_db,C_s)A=C/t_s Per-user origin request rate converts tier throughput into concurrent users. Database fan-out and session capacity then provide independent ceilings.

What each symbol means

N,q,u instances, rated RPS per instance, and target utilization
r,c browser requests/minute and CDN offload fraction
d,Q_db queries per origin request and safe database QPS
C_s,t_s session-store limit and average session minutes

Worked substitution with the default inputs

1. Calculate application supply Q_a=6*280*0.70=1,176 RPSq_u=9/60*(1-0.35)=0.0975 RPS/user Only origin-bound requests consume the application tier.
2. Calculate tier ceilings C_a=floor(1,176/0.0975)=12,061q_db=0.0975*0.6=0.0585 QPS/userC_db=floor(900/0.0585)=15,384 Application and database are independently translated to users.
3. Apply state capacity C=min(12,061,15,384,8,500)=8,500 usersA=8,500/7=1,214.29 starts/min The session store binds the default system.

The default safe capacity is 8,500 active users, 1,300 above the 7,200-user forecast.

Worked situations

Practical examples

  • Nine requests/minute with 35% offload equals 0.0975 origin RPS per user.
  • The default application tier supports 12,061 users, but the 8,500-session store controls capacity.

Better inputs

Useful tips

  • Separate bots and background polling.
  • Measure CDN offload at the same cache policy used in the load test.
  • Test login and checkout endpoints separately when they bypass cache.

Before relying on the result

Limitations and common mistakes

  • One average user behavior cannot represent every page flow.
  • Latency, queues, connection pools, memory, and regional imbalance are not explicit.
  • The session-start estimate assumes steady state and one average duration.

Reference

Key terms

Active user
A user generating the entered request rate during the modeled interval.
Origin request
A request not served by CDN/offload.
Concurrency ceiling
Smallest tier-specific active-user capacity.

Important note

Confirm with production-like load tests that include latency objectives, cache misses, authentication, database connection pools, queues, and failure scenarios.

Frequently asked questions

Why can CDN offload increase user capacity?

Fewer requests reach both the application and its downstream database.

Does average session duration change concurrent capacity?

Not the tier ceilings; it converts concurrency into sustainable session starts per minute.

Can I use page views per month?

No. Convert a peak interval into requests per active user per minute.