Why is a Mysql query execution in Node JS so much slower than a direct Mysql query execution?

Question

Query: select id, event_time from events where event_time > 1395797406712 and event_time < 1398389406712 order by event_time asc.

this query returns ~25k rows (500KB in total size).

When I query the above query in Node.js using the node-mysql driver, it takes ~3-4 seconds to execute. (I used console.time and console.timeEnd)

When I query it directly in mySql, it says it takes ~200 ms.

What accounts for this immense difference, and how do I improve the Node.js implementation to be inline with the direct Mysql query?

Answer 1

Turns out it was an Amazon backend problem. We're using Elastic Beanstalk, and the EC2 and RDS instance are in the same region but not in the same availability zone. Once I set them to be the same, i.e., us-east-1a, the query in node.js took around ~200ms.

Answer 2

Are you using row streaming?

If not, node-mysql is building a 25k element object, with all the allocations and iterations and assignments associated with it. And it's doing it in javascript (rather than nicely optimized native code) since node-mysql is pure javascript.

Row streaming will let you collect the information you need for your calculations one record at a time, much like you would do with a the traditional sort of cursor-based MySQL adapter you see in most other environments.