It is a well-known fact that a fast website is critical towards having high user retention. Google looks favorable upon websites which are well optimized and fast. If you are using a CMS like WordPress or Wix, a lot of optimization is done automatically. If you like to build stuff from scratch like me, there is a ton of work required to optimize a website. This post will cover the 8 things that I did to decrease the load time of this blog written in node by two seconds.
This is the result for a single blog post.
Before the improvements, my home page took 3.14 seconds to load and was 3mb. Now my home page takes 1.22 seconds to load and is only 1.2mb in size. If you look at the waterfall for my home page, most of the time is a result of the youtube embedded videos loading.
Since images are the largest portion of a website's size, optimizing and reducing the size of images will decrease load time. In a perfect web development world, everyone would use SVG images which are extremely small and don't need compression. I wrote a script to automatically optimize JPEG and PNG images for websites since most people don’t use SVG images.
#!/bin/bash
# Simple script for optimizing all images for a website
#
# @author Jeffery Russell 7-19-18
WIDTH="690>" # the ">" tag specifies that images will not get scaled up
folders=("./entries" "./img")
for folder in "${folders[@]}"; do
for f in $(find $folder -name '*.jpg' -or -name '*.JPG'); do
convert "$f" -resize $WIDTH "$f"
jpegoptim --max=80 --strip-all --preserve --totals --all-progressive "$f"
done
for f in $(find $folder -name '*.png' -or -name '*.PNG'); do
convert "$f" -resize $WIDTH "$f"
optipng -o7 -preserve "$f"
done
done
This script will go through the ‘img', and ‘entries’ folders recursively and optimize all the images in there. If an image is more than 690px wide, it will get scaled down. In most cases it is useless to have images with a width greater than 690px because it will just get scaled by the client's web browser.
If you are running a Debian based linux distro, you can download the dependencies for this script with the following commands:
apt-get install jpegoptim
apt-get install optipng
apt-get install convert
The goal of this script is to make most of the images under 100kb for the web. It is ok to have a few images above 100kb; however, you should really avoid having images above 200kb.
One of the largest benefits of Node is its Async abilities: code is executed in a multi-threaded fashion. This can become a "callback hell" if not handled correctly, but, with good code structure it can become very useful and easy to manage. When code is executed in parallel, you can decrease run time by doing other stuff while waiting on costly file IO and database calls.
The problem with Async code is that it is hard to coordinate. Node has a lot of ways to handel synchronization; I prefer to use Promises. Here is a simple example where Async code can be misused.
Bad Async Code:
includes.printHeader(res).then(function()
{
return require(file).main(res, filename, request);
}).then(function()
{
return includes.printFooter(res);
}).catch(function(err)
{
console.log(err);
})
Good Code Async:
Promise.all([includes.printHeader(),
require(file).main(filename, request),
includes.printFooter()]).then(function(content)
{
res.write(content.join(''));
res.end();
}).catch(function(err)
{
console.log(err);
});
In the second example, three blocks of Async code are executed in parallel and in the first example three blocks of async code are executed one after another. Many people may initially do the first option because it may seem like you must create and render the footer after you render the header and body of the page.
A great way to handel async calls is by having most of your methods returning promises which resolve to the HTML or DB information that they produce. When you run Promise.all, it returns an array of the results of the promises which enables you to preserve the order ie header, body, footer. After you do this for all your code, it creates a async tree which runs very fast.
Another Good Async Example:
/**
* Calls blog and sidebar modules to render blog contents in order
*
* @param requestURL
* @returns {Promise|*}
*/
main: function(requestURL)
{
return new Promise(function(resolve, reject)
{
Promise.all([renderPost(requestURL),
require("../sidebar/sidebar.js").main()]).then(function(content)
{
resolve(content.join(''));
}).catch(function(error)
{
reject(error);
})
});
}
Client-side caching is where the client's web browser stores static content they download from your website. For example, if a client caches a CSS style sheet, they won't have to download it again for the next page they visit.
You should cache all images, JavaScript and CSS files since those typically don't change. It is a good idea to set the expiration date of the cache to be something longer than a week, I typically set mine for a month.
For a web browser to accept and cache files, you must set some tags in the HTTP header. In the HTTP header you must specify the content type and cache variables like max age. You also must assign a ETag to the header to give the client a way to verify the content of the cache. This enables the client to detect if there was a change to the file and that they should download it again. Some people set the ETag equal to the version of the stylesheet or JavaScript, but, it is far easier to just set it equal to the hash of the file. I use md5 to hash the files since it is fast and I'm not worried about hash collisions for this application.
You can do this in NGINX if you use it to serve static files, but, you can also do it directly in Node.
var eTag = crypto.createHash('md5').update(content).digest('hex');
result.writeHead(200, {'Content-Type': 'text/css', 'Cache-Control':
'public, max-age=2678400', 'ETag': '"' + eTag + '"',
'Vary': 'Accept-Encoding'});
result.write(content);
result.end();
var eTag = crypto.createHash('md5').update(content).digest('hex');
result.writeHead(200, {'Content-Type': 'image/png',
'Cache-Control': 'public, max-age=2678400',
'ETag': '"' + eTag + '"'});
result.write(content);
result.end();
Even with the best async server, there are still ways to improve performance. If you cache all the static pages that you generate in a HashMap, you can quickly access it for the next web user without ever having to query the database or do file IO.
const cache = require('memory-cache');
var html = cache.get(filename);
if(html == null)
{
// Generate page contents
Promise.all([includes.printHeader(),
require(file).main(filename, request),
includes.printFooter()]).then(function(content)
{
res.write(content.join(''));
res.end();
cache.put(filename, content.join(''));
})
}
else
{
res.write(html);
res.end();
}
I found that it is the fastest to cache everything from static html pages, CSS, JavaScript, and images. For a larger site this may consume a boat load of ram, but, storing images in a HashMap reduces load time since you don't need to read the image file from the disk. For my blog, server-side caching nearly cut my load time in half.
Make sure that you don't accidentally cache a dynamic page like the CMS page in your admin section — really hard to realize while debugging.
To demonstrate the performance increase of this method, I restarted my web server (clearing the cache) and ran a speed test which ran three trials. The first two trials were slow since the server did not have anything in its cache. However, the third trial ran extremely fast since all the contents were in the server's cache.
Compressing content before it is transferred over the internet can significantly decrease the loading time of your website. The only trade off from this approach is that it takes more CPU resources, however, it is well worth it for the performance gains. Using Gzip on CSS and HTML can reduce the size by 60-70%.
If you are running an NGINX server, you can enable Gzip there. There is also a simple node module which will use Gzip compression on an Express app.
npm install compression
var compression = require('compression')
app.use(compression());
If you use a CSS library like Bootstrap or W3-CSS, you will have a ton of css classes which go unused. The standard BootStrap CSS file is around 210kb. After I removed unused CSS definitions the size of the BootStrap file was only 16kb for my website.
For my blog I used the node dependency PurgeCSS to remove unused CSS.
This command will install PurgeCSS for CLI (command line interface).
npm i -g purgecss
This is an example of how you could use PurgeCSS to remove unused css definitions.
purgecss --css css/app.css --content src/index.html --out build/css/
PurgeCSS CLI options.
purgecss --css <css> --content <content> [option]
Options:
--con, --content glob of content files [array]
-c, --config configuration file [string]
-o, --out Filepath directory to write purified css files to [string]
-w, --whitelist List of classes that should not be removed
[array] [default: []]
-h, --help Show help [boolean]
-v, --version Show version number [boolean]
This is not the ideal solution since some CSS definitions may be used on some pages yet unused on other pages. When running this command be sure to select a page which uses all your CSS to prevent losing some CSS styling on certain pages.
You don't have to use this through the command line, you can run this directly in your node app to make it automated. Check out their documentation to learn more.
This is the easiest thing you can do to reduce the size of your website. You just run your CSS and JavaScript through a program which strips out all unnecessary characters.
Ex of Minified CSS:
.bg-primary{background-color:#3B536B!important}#mainNav{font-family:Montserrat,'Helvetica Neue',Helvetica,Arial,sans-serif;font-weight:700;text-transform:uppercase;padding-top:15px;padding-bottom:15px}#mainNav .navbar-nav{letter-spacing:1px}#mainNav .navbar-nav li.nav-item a.nav-link{color:#fff}#mainNav .navbar-nav li.nav-item a.nav-link:hover{color:#D2C0FF;outline:0}#mainNav .navbar-toggler{font-size:14px;padding:11px;text-transform:uppercase;color:#fff;border-color:#fff}.navbar-toggler{padding:.25rem .75rem;font-size:1.09375rem;line-height:1;background-color:transparent;border:1px solid transparent;border-radius:.25rem}.table .thead-dark{color:#fff;background-color:#513E7D;border-color:#32383e}footer{color:#fff}footer h3{margin-bottom:30px}footer .footer-above{padding-top:50px;background-color:#3B536B}footer .footer-col{margin-bottom:50px}footer .footer-below{padding:25px 0;background-color:#3B536B}
There are Node libraries which can minify CSS and Javascript, however, if you are lazy, just use a website like this.
Ignoring the gross amount of Node dependencies you have, it is critical to minimize the amount of dependencies the client needs. I completely removed BootStrap's JavaScript and jQuery from my blog by simply writing a javascript function for my nav bar. This reduced the size of my website by 100kb.
const e = document.querySelector(".navbar-toggler");
const t = document.querySelector(".navbar-collapse");
e.onclick = function()
{
if (e.getAttribute("aria-expanded") == "false")
{
t.classList.remove('collapse');
e.setAttribute('aria-expanded', true);
}
else
{
e.setAttribute("aria-expanded", false);
t.classList.add('collapse');
}
}
You should debate how much you need 3rd party scripts like Google Analytics. In most cases people don't full take advantage of Google Analytics, a simple backend analytics service would work just as good while saving the client load time.