Optimizing InnoDB for high-concurrency REST API traffic

[Sarthak Sarthak]

I am relatively new to advanced MySQL optimization and could use some architecture advice from the veterans here.

I am currently scaling the backend database infrastructure for our platform, LuvSMM, which processes a very high volume of concurrent REST API requests. We are trying to maintain an API response latency of under 200ms globally.

Right now, we are dealing with sudden traffic "surges"—sometimes thousands of concurrent requests hitting our endpoints simultaneously. The workflow is mostly fast SELECT queries (to check user balances), immediately followed by rapid, bulk INSERTs (queuing the new orders).

My questions are:
1. What is the best practice for managing connection pooling in MySQL when dealing with these sudden, massive spikes in API traffic?
2. Are there specific InnoDB settings (like `innodb_buffer_pool_size` or `innodb_flush_log_at_trx_commit`) that I should tweak to prioritize write speed and reduce database-level locking/latency?
3. Is it standard practice to queue these API insertions externally (using something like Redis) before writing to MySQL, or can MySQL handle the raw concurrency if tuned correctly?

I appreciate any guidance or documentation you can point me towards. Thank you for helping a newbie learn the ropes of enterprise-level database scaling!