Load Balancer Capacity Calculator
Precisely Estimate Load Balancer Capacity for Your Applications
Avoid under or over-provisioning your load balancers by estimating the right capacity based on traffic patterns, request rates, and connection counts. This tool helps network engineers and architects plan for scalable, performant, and cost-effective load balancing.
Load Balancer Capacity Calculator
Precisely Estimate Load Balancer Capacity for Your Applications.
About This Tool
The Load Balancer Capacity Calculator is a crucial tool for cloud architects and SREs designing scalable and cost-effective applications. Modern cloud load balancers (like AWS's Application Load Balancer) have complex pricing and capacity models based on multiple dimensions, not just raw bandwidth. This tool demystifies the process by modeling the key factors that determine capacity usage: new connections per second, active concurrent connections, and total processed bytes. It calculates a single 'capacity unit' metric, primarily based on AWS's LCU (Load Balancer Capacity Unit) concept, which provides a standardized way to measure load. By inputting your application's traffic profile, you can get a clear estimate of the required capacity and associated monthly cost. This helps you avoid bottlenecks, prevent dropped connections during traffic spikes, and optimize your cloud networking spend.
How to Use This Tool
- Enter your expected average requests per second for your application.
- Input the number of new connections per second. This is often lower than requests per second if clients reuse connections.
- Provide the peak number of concurrent (active) connections you expect during busy periods.
- Enter the average size of your HTTP requests and responses in kilobytes.
- Select the type of load balancer you are planning to use.
- Click "Calculate Capacity" to see the required number of capacity units and the estimated monthly cost.
- Review the "Dominant Factor" to understand what aspect of your traffic is driving the capacity requirements.
In-Depth Guide
Understanding Load Balancer Capacity Units
Modern cloud load balancers are not priced by the size of the instance they run on, but by the work they do. AWS pioneered this model with the Load Balancer Capacity Unit (LCU). An LCU is an abstract unit that measures the load across four dimensions: **New Connections**, **Active Connections**, **Processed Bytes**, and **Rule Evaluations**. You are billed for the number of LCUs used per hour, based on whichever of these four dimensions is highest. For example, a chat application with long-lived WebSocket connections will be high on 'Active Connections,' while a file download service will be high on 'Processed Bytes.'
New Connections vs. Active Connections
It is important to understand the difference between these two metrics. **New Connections** measures the rate at which new TCP connections are established. This is resource-intensive for the load balancer. **Active Connections** measures the total number of connections that are open at any given moment. For traditional websites, these numbers might be similar. For applications using WebSockets or other long-lived connections, the number of active connections can be very high even if the rate of new connections is low.
The Cost of Rules
Application Load Balancers allow you to create complex rules to route traffic based on hostname, path, headers, etc. Every time the load balancer evaluates these rules, it consumes capacity. While this is often a minor part of the total cost, a very large and complex ruleset can become a significant factor. This is measured in "Rule Evaluations," and 1 LCU provides for 1,000 rule evaluations per second.
Scaling and High Availability
Cloud load balancers are designed to scale automatically in response to traffic. When you provision a load balancer, the provider handles scaling the underlying infrastructure transparently. This means you don't need to worry about "running out" of capacity in the traditional sense. However, the more capacity you use, the higher your bill will be. This calculator helps you predict that bill. For high availability, you should always deploy your load balancer and backend targets across multiple Availability Zones (AZs) within a region.