In high concurrent projects, the cache, the opcode code block, the server, the database configuration, when is it done?

Project is a multi-level concurrency, the amount of data reached tens of millions of levels, the architecture is LNMP, take a redis and opcode.

Like Cache (Redis), database Configuration (MySQL), server configuration (Linux system configuration), are these added to the project during the development process, or will it be reconstructed after the project is completed? How do you weigh this?

My concern is:
If I join the cache (Redis) during the project, do I have a lot of problems during debugging? For example, data synchronization is inconsistent, the code modifies the unresponsive.
But if the project is completed and then reconstructed, the cost seems to be a bit large?

An experienced friend can give a hint. Thank you

Reply content:

Project is a multi-level concurrency, the amount of data reached tens of millions of levels, the architecture is LNMP, take a redis and opcode.

Like Cache (Redis), database Configuration (MySQL), server configuration (Linux system configuration), are these added to the project during the development process, or will it be reconstructed after the project is completed? How do you weigh this?

My concern is:
If I join the cache (Redis) during the project, do I have a lot of problems during debugging? For example, data synchronization is inconsistent, the code modifies the unresponsive.
But if the project is completed and then reconstructed, the cost seems to be a bit large?

An experienced friend can give a hint. Thank you

My advice is: the production environment and development environment should be consistent, to avoid some problems on the line to appear.
For configuration, you can configure an intranet and an external network.
You can deploy the intranet redis,mysql, and then write a copy of the intranet configuration, to the online after the configuration of the external network.
If you don't do it, it's dangerous to change the configuration on the wire.

