Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.Original topic: DM同步延迟严重，怎么加快同步速度
|
username: Hacker_PtIIxHC1
Using DMv2.0 to synchronize data from MySQL to TiDB (v4.0.8), due to the large amount of upstream changes, the downstream TiDB experiences significant delays. How can the synchronization speed be improved?
The configuration file for the task has the following parameters, which can be appropriately increased based on the hardware:
syncers:
global:
worker-count: 16
batch: 100
The documentation has a troubleshooting approach, can we refer to it?
It seems that your version is very low. You can check if there are any configuration options for threads.
After adjusting these two parameters, it feels like there’s no change. 
Check your server resources, such as CPU, memory, and I/O. Are they reaching their limits?
Are you using incremental or full + incremental? If the data volume is relatively large, it’s better to use dumpling + lightning for full and DM for incremental. DM can handle up to 30~50 GiB/h at most.
What is the scenario, and what kind of business is typically upstream?
Identify the bottleneck. If there are no issues with IO, CPU, memory, or physical resources, increase the number of threads.
We need to consider network I/O, physical resources, and other issues from multiple aspects.
Refer to the performance test report:
Additionally, it seems that your DM version is a bit outdated; you might want to upgrade it.
Sure, please provide the Chinese text you need translated into English.
Make sure that IO, CPU, memory, and physical resources are not an issue before increasing the number of threads.
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: 
|
username: