Tencent Cloud

Recent Pages

Use Limits

Last updated: 2024-11-25 16:42:14

This document outlines some limitations on metrics and performance in TDMQ for CKafka. Be mindful not to exceed these limits during use to avoid any issues.
Item
Description
Number of topics
The topic limit depends on product specifications. For details, see Billing Overview | Tencent Cloud.
Number of partitions
One topic can support up to 3,000 partitions.
The instance-level partition limit includes replicas, which are typically 2 or 3.
Reducing the number of partitions is not supported.
Partition throughput
In the case of ack = 1, the throughput of a single TDMQ for CKafka partition is between 30 and 60 MB/sec due to factors such as TDMQ for CKafka's partition architecture, business data size, and request frequency.
When using ack = -1 (strong consistency), we recommend maintaining the throughput of a single TDMQ for CKafka partition between 10 and 20 MB/sec. This helps ensure that the stability of request duration is not impacted by factors such as TDMQ for CKafka's partition architecture, business data size, and request frequency.
In scenarios involving idempotency or transactions, CKafka server load may increase, impacting the maximum throughput per partition. If your business is latency-sensitive, it is recommended to test with different cluster specifications.
Duration
TDMQ for CKafka is a high-throughput, high-volume message queue and cannot guarantee low latency for every request. It is recommended to set the timeout as follows:
For the producer, when ACK = 1, the timeout period defaults to 30s.
For the producer, when ACK = -1, the timeout period defaults to 60s.
The timeout period for the consumer is set to 60s.
Number of consumer groups
For Standard Edition, the recommended number of instance-level consumer groups is up to 50.
For Advanced and Professional Editions, it is recommended that the number of consumer groups per instance not exceed 200. Contact Us for upgrade options.
Client Usage
It is recommended to use the CKafka control console or TencentCloud API for managing clusters, topics, and other metadata.
It is not recommended to directly use AdminClient to create topics, modify partitions, or perform other metadata operations to avoid data inconsistency in the console. Such actions are planned to be blocked in the future.
Instance
The region attribute of instances cannot be changed.
The maximum number of client connections to a Standard Edition instance is 5,000, and to a Pro Edition instance, 50,000. When the limit is reached, the client cannot create more connections. If this maximum value is unreasonable for your actual business, you can submit a ticket to request an increase.
Version
Compatible with open-source versions 2.4, 2.8, and 3.2
Routes
An instance can create up to five routes, with only one of them being a public network route.
Public network bandwidth
Not supported.
Exposing ZooKeeper
Not supported.
Exposing underlying resources
It is not supported so as to avoid risks caused by user's operation.
Message size
It cannot exceed 12 MB; otherwise, messages will fail to be sent.
Tag
Each Tencent Cloud resource can have up to 50 tags.
Concurrent operations in the console
To prevent consistency issues from high concurrency in instance metadata operations, some interfaces are locked to limit requests.
Currently, only 20 concurrent console requests per instance are allowed to ensure stability and success (including SDK API calls).
Note: 
Due to TDMQ for CKafka's partition architecture, business data size, request frequency, and the stability of the physical layer, TDMQ for CKafka cannot guarantee that each request has low latency.
However, we strive to ensure the following two requirements:
The percentage of low-latency production for a single instance each month aligns with the instance's SLA.
The percentage of low-latency consumption for a single instance each month aligns with the instance's SLA.
The calculation formulas are as follows:
The percentage of low-latency production for a month = (Number of minutes in the month when 99.9th percentile latency production time per minute is 30 seconds or less / Total service minutes in the month) × 100%.
The percentage of low-latency consumption for a month = (Number of minutes in the month when 99.9th percentile latency consumption is 60 seconds or less / Total service minutes in the month) × 100%.
﻿

Contact Us

Contact our sales team or business advisors to help your business.

Technical Support

Open a ticket if you're looking for further assistance. Our Ticket is 7x24 avaliable.

7x24 Phone Support

tencent cloud

Recent Pages

Use Limits

Was this page helpful?

Was this page helpful?

Item	Description
Number of topics	The topic limit depends on product specifications. For details, see Billing Overview \| Tencent Cloud.
Number of partitions	One topic can support up to 3,000 partitions. The instance-level partition limit includes replicas, which are typically 2 or 3. Reducing the number of partitions is not supported.
Partition throughput	In the case of `ack = 1`, the throughput of a single TDMQ for CKafka partition is between 30 and 60 MB/sec due to factors such as TDMQ for CKafka's partition architecture, business data size, and request frequency. When using `ack = -1` (strong consistency), we recommend maintaining the throughput of a single TDMQ for CKafka partition between 10 and 20 MB/sec. This helps ensure that the stability of request duration is not impacted by factors such as TDMQ for CKafka's partition architecture, business data size, and request frequency. In scenarios involving idempotency or transactions, CKafka server load may increase, impacting the maximum throughput per partition. If your business is latency-sensitive, it is recommended to test with different cluster specifications.
Duration	TDMQ for CKafka is a high-throughput, high-volume message queue and cannot guarantee low latency for every request. It is recommended to set the timeout as follows: For the producer, when ACK = 1, the timeout period defaults to 30s. For the producer, when ACK = -1, the timeout period defaults to 60s. The timeout period for the consumer is set to 60s.
Number of consumer groups	For Standard Edition, the recommended number of instance-level consumer groups is up to 50. For Advanced and Professional Editions, it is recommended that the number of consumer groups per instance not exceed 200. Contact Us for upgrade options.
Client Usage	It is recommended to use the CKafka control console or TencentCloud API for managing clusters, topics, and other metadata. It is not recommended to directly use AdminClient to create topics, modify partitions, or perform other metadata operations to avoid data inconsistency in the console. Such actions are planned to be blocked in the future.
Instance	The region attribute of instances cannot be changed. The maximum number of client connections to a Standard Edition instance is 5,000, and to a Pro Edition instance, 50,000. When the limit is reached, the client cannot create more connections. If this maximum value is unreasonable for your actual business, you can submit a ticket to request an increase.
Version	Compatible with open-source versions 2.4, 2.8, and 3.2
Routes	An instance can create up to five routes, with only one of them being a public network route.
Public network bandwidth	Not supported.
Exposing ZooKeeper	Not supported.
Exposing underlying resources	It is not supported so as to avoid risks caused by user's operation.
Message size	It cannot exceed 12 MB; otherwise, messages will fail to be sent.
Tag	Each Tencent Cloud resource can have up to 50 tags.
Concurrent operations in the console	To prevent consistency issues from high concurrency in instance metadata operations, some interfaces are locked to limit requests. Currently, only 20 concurrent console requests per instance are allowed to ensure stability and success (including SDK API calls).

tencent cloud

Sign Up

Log in

Recent Pages

Use Limits

Was this page helpful?

Was this page helpful?