CODEWITHSUNDEEP
node.jsmongodb
Discipol
Discipol

Reputation: 3157

Should I go for faster queries or less cpu consuption / faster processing?

I have to choose between performing a query for X size data and not process it, just send it to the client,

OR

I can choose to perform a query for half X size data and do a little processing, then send it to the client.

Now, in my life of a programmer I met storage vs speed problem quite a bit, but in this case, I have to choose between "fast query + processing" or "slow query + no processing".

If it matters, I am using nodejs for the server and mongodb for the database. If you care, I am holding non intersecting map areas and I am testing if an area intersects any or no map area. All are boxes. If I keep them as origin point, its only one pair of coordinates and I have to process the point into an area(all map areas have the same size). If I store them as an area directly, I don't have to process it anymore, but its 4 pairs of coordinates now. 4 times the size and I think, slower query.

Upvotes: 0

Views: 84

Answers (1)

Niels Keurentjes
Niels Keurentjes

Reputation: 41968

There is no right answer to this question, it all depends on your infrastructure. If you're for example using Amazon Webservices for this, it depends on the transaction price. If you've got your own infrastructure, it depends on the load of the DB and web servers. If they are on the same server, it's a matter of the underlying hardware whether the I/O from the DB starts to limit before the CPU/memory become the bottle neck.

The only way to determine the right answer to this question for your specific situation is to set it up and do a stress test, for example using Load Impact or one of the tons of other good tools to do this. While it is getting hammered, monitor your system load using top and watch the wa column specifically - if it starts going up over 50% consistently you're I/O limited, and the DB should be offloaded to the CPU.

Upvotes: 4

Related Questions