Key metrics for measuring Web performance

Since the birth of the site, response speed/response time has always been the topic of concern, and slow is a site of the killer, just as we think that the four-core and broadband capabilities to solve these problems, Wi-Fi and mobile devices for hot mobile internet and quietly rise.

In 2006, Amazon reported a 1% increase in response time per 100ms. The value of optimization is obvious, but how fast is a standard, or how fast is fast enough? So what exactly is response time, and how much value does it have?

Technically, response time refers to the time a user sends a command (for example, a page request) when the browser receives the completed load. The definition looks very simple, but when you think about how to design a modern web page with many additional objects, the response time is very important to the user experience, and it does not tell you which factors affect the response time.

A slightly better metric is page load time. Page load time refers to the time taken from the user sending instructions to the browser to load the entire Page object. Like response time, the page loads the whole process involves a lot of things, it consists of a series of execution steps, and each step needs to be monitored separately, each step will tell you where the problem lies.

The steps include:

• DNS resolution TIME
• TCP Link Time
• HTTP REDIRECT Time
• First byte load time
• HTML content Time
• Entire Page object load time

　　DNS resolution TIME

DNS lookup time is to translate the domain name into a specific IP time, most people think that whether or not DNS work, is not a simple thing.

In this process, you may encounter a number of subtle problems, such as long response times, timeouts, invalid caches, and so on. In these cases, a query can pass, but it takes more time.

In general, if DNS lookup time is too long, it means that you or the hosting service has a problem with the DNS service. Keep in mind that if the site is too far away from its DNS service, the parsing time will be slightly increased, which will be reflected on some international websites, while a valid cache will reduce the time.

　　TCP Link Time

When the URL is resolved to an IP address, the TCP link time represents the amount of time that the client has spent linking to the server. Monitoring link time helps developers to identify issues that affect response time, such as network latency, routing problems, and server broadband issues.

For example, if the broadband server is not sufficient to handle the workload, then the client is aware of the problem with the server side, and when the client sends the request to the server, it may be denied or time-out, response time delay, and so on.

　　HTTP REDIRECT Time

The HTTP redirect time is primarily the time that the TCP link finishes, which means the time it takes to send an initial notification to the redirected Web site and ultimately the browser is directed to the target site. If there is no redirection, then the redirect time is 0. It includes DNS parsing time, TCP links, and so on.

HTTP redirection can be used to shorten URLs, when a page link moves, to prevent link corruption, or to allow multiple domain names to be linked to a Web site.

　　First byte load time

When developers think about how to optimize a site, they often choose to optimize the content-file combination, multimedia optimization, caching, and compressing files-but there is also a need to optimize the server. One of the best indicators is the load time of the first byte, and the first byte load time represents the time taken from the link creation to the first byte of the successful conversion. This time also includes the time the server performs various protocols and calculations.

Typically, the server side encounters problems related to the first byte, including memory leaks, too many process-derived processes-not completely shut down-inefficient SQL queries, and calls to external resources such as Google and Facebook.

　　HTML content Time

HTML content time mainly includes loading the Web page layout, CSS, JavaScript, this time and the size of the HTML page has a direct relationship. HTML content load times are often used as a metric for measuring broadband, but not entirely.

　　Entire Page object load time

Once the entire HTML content is received, the browser resolves all the page objects and until all objects have been loaded. These objects include images, JavaScript, CSS, Flas objects, RSS feeds, JavaScript files, and so on.

Measuring full-page load times is useful for monitoring third-party content, especially ads, but it does not tell you which users have seen the ad. For example, it doesn't tell you where third-party content is loaded faster. But from the user's point of view, these are not problems.

The author makes a detailed division and explanation of the response time and page load time of the website, I believe you have a deeper understanding of these indicators, developers can based on these indicators to determine the site of the problem (if there is a problem).

Key metrics for measuring Web performance