Experiments

Addressing and Analyzing the problems faced by websites affecting their web performance.

Case Study:- Facebook's web performance
Three main parts focussing web performance:-

1. Network Time Five main aspects:- Bytes of cookies, HTML, CSS, JS, images

2. Generation Time a. Client sends request. The request hits the web server to get the response from the web server. So, the capture of time between the (i) sender and receiver (ii) receiver and sender is termed as generation time. b. Metrics:- Code efficiency, web server response time, caching, database, network hardware. c. Goal:- Reduce generation time through cleaner, faster code and improve backend architectures.

3. Render Time a. Measuring the time of web browser between the client's request and response from website's web server and displaying the destined web page. b. Performance and behaviour of the web browser is focussed. c. Minimizing bytes of HTML, CSS, Javascript and Images helps in rendering time. d. Javascript execution.

Additional:- a. Metrics: TTI (Time-To-Interact) measures the time taken for the newsfeed to get visible.

The Bottom Line is:- Inspite of these measures, yet Facebook is slow in web performance.

Experimental demonstration:-

Experiment 1:
Technology used:- Cygwin integrated with node (requires internet)



Experiment 2:
Technology used:- Google Cloud - BigQuery (requires internet)

A test query for websites:-

SELECT respSize, respCookieLen, respHeadersSize, respHttpVersion, resp_vary, _cdn_provider, resp_cache_control, resp_content_encoding, _gzip_save, resp_transfer_encoding, resp_age, resp_connection, resp_accept_ranges, resp_server, urlShort 
FROM [httparchive:runs.latest_requests] 
LIMIT 10

Experiment:

SELECT requestid, urlShort, reqHttpVersion, reqHeadersSize, respSize, mimeType, respCookieLen, resp_cache_control, _gzip_save, resp_server
FROM [httparchive:runs.2014_11_15_requests_mobile] 
LIMIT 10