This topic describes cluster limits, capacity limits, and concurrency limits for Alibaba Cloud Container Compute Service (ACS), and explains how to request higher quotas.
Overview of limits
Before using Alibaba Cloud Container Compute Service (ACS), note the following usage limits:
After creating an ACS cluster, you cannot perform the following operations:
Change the VPC of the cluster.
Change the container network plugin.
Change the storage plugin.
Migrate applications across namespaces.
Traffic to cluster control plane components is subject to the following limits:
When you access cluster control plane components (API Server and etcd) through APIs or the command line, rate limiting may trigger if you read many cluster events in a single request due to bandwidth constraints, causing the read operation to fail. We recommend that you use the ACS console Operations Management > Event Hub to query cluster events, or add pagination parameters in your API or command-line requests to reduce the amount of data per request (for example:
--chunk-size=500).
Quotas
Type | Quota limit | Default quota | How to increase quota |
Total number of ACS clusters that a single account can create | 20 | 3 | Go to the Quota Center to request a higher default quota. Select the quota named Total ACS Clusters. Note If the quota limit still does not meet your needs, submit a ticket to contact technical support for a higher quota. Select the product category Container Service for Kubernetes. |
Maximum total vCPUs available for pay-as-you-go elastic instances in a single region | 30,000 vCPUs | 100 vCPUs | Go to the Quota Center and select the target region to request a higher default quota. Note If the quota limit still does not meet your needs, submit a ticket to contact technical support for a higher quota. Select the product category Container Compute Service. |
Capacity limits
Type | Quota limit | How to increase quota |
etcd storage capacity | 8 GB | Cannot be increased |
Total size of each type of etcd object | 800 MB | Cannot be increased |
Pod | 50,000 | Requests exceeding 50,000 pods are rejected by default. submit a ticket to request technical support for a higher quota. Important If many pods are associated with services, keep the number below 20,000. |
service | 10,000 units | Requests exceeding 10,000 Services are rejected by default. submit a ticket to request technical support for a higher quota. Note If a namespace contains many Services, the large number of environment variables injected by kubelet may slow down or prevent Pod startup. To resolve this performance bottleneck, set |
ConfigMap | 30,000 units | submit a ticket to request technical support for a higher quota. |
Secret | 10,000 units | |
PVC | 10,000 units | |
PV | 10,000 units | |
Total CRs of a single type | 100,000 | The default resource limit is 10,000. Requests exceeding this limit are rejected. submit a ticket to request technical support for a higher quota. |
Total CRDs | 100,000 units | Cannot be increased |
Concurrency limits
Cluster control plane
QPS (Queries Per Second) refers to the number of requests received per second.
Type | Concurrency limit | Default concurrency | How to increase concurrency |
Full list QPS for Pods | 1 | 1 | Frequent full retrieval of core Pod data puts heavy pressure on the control plane. Optimize your code by adding |
Pod Create QPS | 300 | 300 | Requests exceeding 300 QPS are rejected by default and return HTTP status code 429. submit a ticket to request technical support for higher concurrency. |
Pod Delete QPS | 300 | 300 | |
Pod Patch QPS | 300 | 300 | |
Pod Update QPS | 300 | 300 | |
Services Create QPS | 20 | 20 | Requests exceeding 20 QPS are rejected by default and return HTTP status code 429. submit a ticket to request technical support for higher concurrency. |
Services Delete QPS | 30 | 30 | Requests exceeding 30 QPS are rejected by default and return HTTP status code 429. submit a ticket to request technical support for higher concurrency. |
Services Patch QPS | 30 | 30 | |
Services Update QPS | 30 | 30 | |
Lease Update QPS | 2000 | 2000 | Requests exceeding 2000 QPS are rejected by default and return HTTP status code 429. submit a ticket to request technical support for higher concurrency. |
ConfigMap Update QPS | 200 | 200 | Requests exceeding 200 QPS are rejected by default and return HTTP status code 429. submit a ticket to request technical support for higher concurrency. |
Endpoint Update QPS | 200 | 200 | |
Other Create QPS | 500 | 500 | Requests exceeding 500 QPS are rejected by default and return HTTP status code 429. submit a ticket to request technical support for higher concurrency. |
Other Delete QPS | 500 | 500 | |
Other Patch QPS | 500 | 500 | |
Other Update QPS | 500 | 500 |
Image cache
The following table lists the maximum number of API calls per minute for each API in a single region.
API operation | Maximum calls per minute |
CreateImageCache | 300 |
DeleteImageCache | 300 |
ListImageCaches | 600 |
GetImageCache | 6000 |
Instance delivery throughput
The following table lists the maximum Pod delivery throughput per minute for instance-related operations in a single region (including internal retries).
Instance-related capability | Maximum Pod delivery throughput per minute |
Instance creation | 1000 |
Instance deletion | 1000 |
Quota limits of underlying cloud services
Limit category | Limitations | Default quota | How to increase quota |
Custom route entries per route table (excluding dynamically propagated route entries). | 200 | Go to the quota management page or the Quota Center to request a higher quota. | |
Dynamically propagated routes per table. | 500 | ||
vSwitches per VPC. | 150 | ||
VPCs in a single region. | 10 | ||
Private IP addresses used by cloud resources in a VPC. | 300,000 1. If an Elastic Compute Service (ECS) instance has only one private IP, the ECS instance uses only one network address. 2. If an ECS instance is associated with multiple ENIs or multiple IP addresses are configured for the ENIs, the number of network addresses used by the ECS instance is the sum of the IP addresses that are assigned to the ENIs associated with the ECS instance. | No. | |
Maximum number of private IP addresses that a security group in a VPC can contain per Alibaba Cloud account in a region |
| No. | |
Maximum number of ENIs (secondary ENIs) that a single Alibaba Cloud account can create in a specific region | View in the Quota Center. | Go to the Quota Center to increase the quota. | |
Pay-as-you-go EIPs per account (excludes subscription EIPs) | 20 | Go to the Quota Center to increase the quota. | |
Number of CLB instances that can be created per Alibaba Cloud account | 30 | You can increase the quota using one of the following methods:
| |
Number of backend servers that can be added to a CLB instance | 200 | You can increase the quota using one of the following methods:
| |
Number of listeners that can be added to a CLB instance | 50 | You can increase the quota using one of the following methods:
| |
Quota for the number of pay-as-you-go disks across all regions for a single account | View in the Quota Center. | submit a ticket to increase the quota. | |
Quota for the capacity of pay-as-you-go disks used as data disks for a single account | Depends on your ECS usage, region, and disk type. View in the Quota Center. For more information, see View or increase Elastic Block Storage quotas. | submit a ticket to increase the quota. |