GO_Cache

Steps to run

1. cd src/github.com/ParallelDots/server and then go build server.go
2. cd src/github.com/ParallelDots/client and then go build client.go
3. The above steps will create two executables server or server.exe(if on Windows) and client or client.exe.
4. Run server and client executables in seperate terminals.
5. For server use server -cache to enable caching as by default it is disabled and for client use client -requests=value to specify number of concurrent requests to send.

Approach

I thought of a hybrid approach where cache can live as a persistent layer as well as on RAM. To make it persistent I first thought of saving the data on a file after a certain period of time which would require event handling if the server is still serving up requests. So, instead I implemented it in such a way that when the server is shutdown the whole data of the cache is saved into a file, and when next time the server is spawned it can load that file and use the data stored to serve requests.

I used maps to store data as they provide constant lookup time. Also I used gob package to encode and decode data as the data will reside in RAM as well as hard disk.

There are many improvements that can be made like adding a timer to cache so that the unwanted/least used data can be removed from cache, serving multiple servers at once(that will require more synchronization) , etc.

Benchmarking Results

1. Without caching for 100 requests each request takes around 30.75 - 30.95 seconds.
2. With caching enabled for 80 requests the time is 54-70 milli seconds and for the remaining 20 requests it is between 30.077 - 30.089 seconds.

We see a significant improvement here from ~30 seconds to 54 milli seconds for cached requests.