Request throttling for the Elastic Load Balancing API
Elastic Load Balancing throttles its API requests for each AWS account on a per-Region basis. We do this to help the performance and availability of the service. Throttling ensures that requests to the Elastic Load Balancing API do not exceed the maximum allowed API request limits. API requests are subject to the request limits whether you call them or they are called on your behalf (for example, by the AWS Management Console or a third-party application).
If you exceed an Elastic Load Balancing API throttling limit, you get the ThrottlingException
error code and a Rate exceeded error message.
We recommend that you prepare to handle throttling gracefully. For more information, see
Timeouts, retries, and backoff with jitter
How throttling is applied
Elastic Load Balancing uses the token bucket
algorithm
Elastic Load Balancing provides two sets of API actions. ELB API version 2 supports the following types of load balancers: Application Load Balancers, Network Load Balancers, and Gateway Load Balancers. ELB API version 1 supports Classic Load Balancers. Each ELB API version has its own buckets and tokens.
Services that call the Elastic Load Balancing API on your behalf, such as Amazon EC2, Amazon ECS, Amazon EC2 Auto Scaling, and AWS CloudFormation have their own account-level buckets. These services do not consume tokens from your buckets.
Request rate limiting
With request rate limiting, you are throttled on the number of API requests that you
make. Each request that you make removes one token from the bucket. For example, the
token bucket size for non-mutating API actions is 40 tokens. You can make up to 40
Describe* requests in one second. If you exceed 40
Describe* requests in one second, you are throttled and the remaining
requests within that second fail.
Buckets automatically refill at a set rate. If a bucket is below its maximum
capacity, a set number of tokens is added back every second until the bucket reaches its
maximum capacity. If a bucket is full when refill tokens arrive, they are discarded.
A bucket can't hold more than its maximum number of tokens. For example, the bucket
size for non-mutating API actions is 40 tokens and the refill rate is 10 tokens per
second. If you make 40 DescribeLoadBalancers requests in one second, the
bucket is reduced to zero (0) tokens. We add 10 refill tokens to the bucket every
second, until it reaches its maximum capacity of 40 tokens. This means that it takes
4 seconds for an empty bucket to reach its maximum capacity, if no requests are
made during that time.
You do not need to wait for a bucket to be completely full before you can make API requests. You can use tokens as they are added to a bucket. If you immediately use the refill tokens, the bucket does not reach its maximum capacity.
There is an account-level throttling limit that is shared across all Elastic Load Balancing API actions. The capacity of the account-level bucket is 40 tokens and the refill rate is 10 request tokens per second.
Request token bucket sizes and refill rates
For request rate limiting purposes, API actions are grouped into categories. Each category has its own limits.
Categories
-
Mutating actions — API actions that create, modify, or delete resources. This category generally includes all API actions that are not categorized as non-mutating actions. These actions have a lower throttling limit than non-mutating API actions.
-
Non-mutating actions — API actions that retrieve data about resources. These API actions typically have the highest API throttling limits.
-
Resource-intensive actions — API actions that take the most time and consume the most resources to complete. These actions have an even lower throttling limit than mutating actions. These actions are throttled separately from other mutating actions.
-
Registration actions — API actions that register or deregister targets. These API actions are throttled separately from other mutating actions.
-
Uncategorized actions — These API actions receive their own token bucket sizes and refill rates, even though they fall under one of the other categories.
The following table shows the default capacity and refill rates for the categorized request token buckets.
| Category | ELBv2 actions | ELBv1 actions | Bucket capacity | Refill rate (per second) |
|---|---|---|---|---|
| Resource-intensive | CreateLoadBalancer, SetSubnets |
CreateLoadBalancer,
AttachLoadBalancerToSubnets,
DetachLoadBalancerFromSubnets,
EnableAvailabilityZonesForLoadBalancer,
DisableAvailabilityZonesForLoadBalancer |
10 | 0.2 † |
| Registration | RegisterTargets, DeregisterTargets |
RegisterInstancesWithLoadBalancer,
DeregisterInstancesFromLoadBalancer |
20 | 4 |
| Non-mutating | DescribeAccountLimits,
DescribeCapacityReservation,
DescribeListenerAttributes,
DescribeListenerCertificates,
DescribeListeners,
DescribeLoadBalancerAttributes,
DescribeLoadBalancers, DescribeRules,
DescribeSSLPolicies, DescribeTags,
DescribeTargetGroupAttributes,
DescribeTargetGroups,
DescribeTargetHealth |
Describe* |
40 | 10 |
| Mutating | AddListenerCertificates, AddTags,
CreateListener, CreateRule,
CreateTargetGroup, DeleteListener,
DeleteLoadBalancer, DeleteRule,
DeleteTargetGroup,
ModifyCapacityReservation, ModifyIpPools,
ModifyListener, ModifyListenerAttributes,
ModifyLoadBalancerAttributes, ModifyRule,
ModifyTargetGroup,
ModifyTargetGroupAttributes,
RemoveListenerCertificates, RemoveTags,
SetIpAddressType, SetRulePriorities,
SetSecurityGroups |
AddTags, ApplySecurityGroupsToLoadBalancer,
ConfigureHealthCheck,
CreateAppCookieStickinessPolicy,
CreateLbCookieStickinessPolicy,
CreateLoadBalancerListener,
CreateLoadBalancerPolicy, Delete*,
ModifyLoadBalancerAttributes, RemoveTags,
SetLoadBalancer* |
20 | 3 |
The following table shows the default capacity and refill rates for the uncategorized request token buckets for ELBv2.
| ELBv2 actions | Bucket capacity | Refill rate (per second) |
|---|---|---|
CreateTrustStore |
10 | 0.2 † |
AddTrustStoreRevocations,
DeleteSharedTrustStoreAssociation,
DeleteTrustStore, ModifyTrustStore,
RemoveTrustStoreRevocations |
10 | 0.2 † |
GetResourcePolicy,
GetTrustStoreCaCertificatesBundle,
GetTrustStoreRevocationContent |
20 | 4 |
DescribeTrustStoreAssociations,
DescribeTrustStoreRevocations,
DescribeTrustStores |
40 | 10 |
† Fractional refill rates require several seconds to generate one full token.
Monitoring API requests
You can use AWS CloudTrail to monitor your Elastic Load Balancing API requests. For more information, see Log API calls for Elastic Load Balancing using AWS CloudTrail.
