IoT Hub
IoT Hub Documentation
Getting Started Documentation Guides API FAQ

API & Rate Limits

API & Rate Limits feature allows controlling API usage, by limiting number of requests from a single host/device/tenant during single time unit (Minutes, Hours, etc.).

API & Rate limits are disabled by default. System administrator is able to configure rate limits using thingsboard.yml.

REST API limits

REST API calls are used by all sorts of UI components and possibly some automatic scripts launched on behalf of customer user or tenant user. It is critical to limit amount of API calls by tenant or by customer to avoid overloading the server due to mistakes in a custom widget or script.

The rest.limits.tenant.enabled parameter or TB_SERVER_REST_LIMITS_TENANT_ENABLED environment property enables/disables tenant level limits.

The rest.limits.tenant.configuration parameter or TB_SERVER_REST_LIMITS_TENANT_CONFIGURATION environment property configure maximum amount of REST API calls. For example, value “100:1,2000:60” means no more then 100 requests per second and no more then 2000 requests per minute.

1
2
3
4
5
6
7
8
9
10

server:
  ...
  rest:
    limits:
      tenant:
        enabled: "${TB_SERVER_REST_LIMITS_TENANT_ENABLED:false}"
        configuration: "${TB_SERVER_REST_LIMITS_TENANT_CONFIGURATION:100:1,2000:60}"
      customer:
        enabled: "${TB_SERVER_REST_LIMITS_CUSTOMER_ENABLED:false}"
        configuration: "${TB_SERVER_REST_LIMITS_CUSTOMER_CONFIGURATION:50:1,1000:60}"

Websocket limits

Websockets are used to deliver real-time notifications about new telemetry values from device to the dashboard.

The ws.send_timeout parameter or TB_SERVER_WS_SEND_TIMEOUT environment property controls maximum time for a successful websocket message delivery to the client. If client is too slow, the session will be closed.

The ws.limits.max_queue_per_ws_session parameter or TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_QUEUE_PER_WS_SESSION environment property controls max messages that are awaiting delivery to the client. If client is too slow, the session will be closed.

The ws.limits.max_sessions_per_* parameter or TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SESSIONS_PER_* environment property controls maximum amount of active connections per certain entity: tenant, customer, public or regular user. If there is too much sessions per certain criteria, new connections will be dropped.

The ws.limits.max_subscriptions_per_* parameter or TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SUBSCRIPTIONS_PER_* environment property controls maximum amount of active subscriptions within all sessions per certain entity: tenant, customer, public or regular user. If there is too much subscriptions per certain criteria, new subscriptions will not be accepted.

The ws.limits.max_updates_per_session parameter or TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_UPDATES_PER_SESSION environment property configure maximum amount of messages sent to the client from the server for each session. For example, value “300:1,3000:60” means no more then 300 updates per second and no more then 3000 updates per minute.

You can find sample configuration below:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

server:
  ...
  ws:
    send_timeout: "${TB_SERVER_WS_SEND_TIMEOUT:5000}"
    limits:
      # Limit the amount of sessions and subscriptions available on each server. Put values to zero to disable particular limitation
      max_sessions_per_tenant: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SESSIONS_PER_TENANT:0}"
      max_sessions_per_customer: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SESSIONS_PER_CUSTOMER:0}"
      max_sessions_per_regular_user: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SESSIONS_PER_REGULAR_USER:0}"
      max_sessions_per_public_user: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SESSIONS_PER_PUBLIC_USER:0}"
      max_queue_per_ws_session: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_QUEUE_PER_WS_SESSION:500}"
      max_subscriptions_per_tenant: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SUBSCRIPTIONS_PER_TENANT:0}"
      max_subscriptions_per_customer: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SUBSCRIPTIONS_PER_CUSTOMER:0}"
      max_subscriptions_per_regular_user: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SUBSCRIPTIONS_PER_REGULAR_USER:0}"
      max_subscriptions_per_public_user: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_SUBSCRIPTIONS_PER_PUBLIC_USER:0}"
      max_updates_per_session: "${TB_SERVER_WS_TENANT_RATE_LIMITS_MAX_UPDATES_PER_SESSION:300:1,3000:60}"

Database rate limits

Although we are limiting users by amount of REST API calls, some of the calls may produce more then one database query. Similar, rule chains may also cause a lot of queries during the message processing. Single telemetry upload also causes database queries to write the data to DB.

You can specify the limits for Cassandra DB and you can also specify parameters to print the statistics. For example:

2018-11-29 10:51:25,020 [SockJS-1] INFO  o.t.s.d.n.CassandraBufferedRateExecutor - 
Permits queueSize [0] totalAdded [6395] totalLaunched [6395] totalReleased [6396] totalFailed [0] totalExpired [0] 
totalRejected [0] totalRateLimited [0] totalRateLimitedTenants [0] currBuffer [0]

Optionally, you can specify to print the tenant names that are violating the thresholds.

The cassandra.query.tenant_rate_limits.configuration parameter or CASSANDRA_QUERY_TENANT_RATE_LIMITS_CONFIGURATION environment property configure maximum amount of queries per tenant. For example, value “1000:1,30000:60” means no more then 1000 updates per second and no more then 30000 updates per minute.

1
2
3
4
5
6
7
8
9
10
11

# Cassandra driver configuration parameters
cassandra:
  ...
  # Cassandra cluster connection query parameters
  query:
  ...
    rate_limit_print_interval_ms: "${CASSANDRA_QUERY_RATE_LIMIT_PRINT_MS:10000}"
    tenant_rate_limits:
      enabled: "${CASSANDRA_QUERY_TENANT_RATE_LIMITS_ENABLED:false}"
      configuration: "${CASSANDRA_QUERY_TENANT_RATE_LIMITS_CONFIGURATION:1000:1,30000:60}"
      print_tenant_names: "${CASSANDRA_QUERY_TENANT_RATE_LIMITS_PRINT_TENANT_NAMES:false}"

Transport rate limits

It is quite important to be able to limit amount of messages that are accepted from a single device or from all devices per tenant. This limit is applied on a transport level, before messages are pushed to the rule engine.

The transport.rate_limits.tenant.configuration parameter or TB_TRANSPORT_RATE_LIMITS_TENANT environment property configure maximum amount of messages from all devices per tenant. For example, value “1000:1,20000:60” means no more then 1000 messages per second and no more then 20000 updates per minute.

The transport.rate_limits.tenant.configuration parameter or TB_TRANSPORT_RATE_LIMITS_DEVICE environment property configure maximum amount of messages from single device. For example, value “10:1,300:60” means no more then 10 messages per second and no more then 300 updates per minute.

1
2
3
4
5
6

transport:
...
  rate_limits:
    enabled: "${TB_TRANSPORT_RATE_LIMITS_ENABLED:false}"
    tenant: "${TB_TRANSPORT_RATE_LIMITS_TENANT:1000:1,20000:60}"
    device: "${TB_TRANSPORT_RATE_LIMITS_DEVICE:10:1,300:60}"

Next steps

  • Getting started guides - These guides provide quick overview of main IoT Hub features. Designed to be completed in 15-30 minutes.