As a hosting provider, our priority is making sure your sites are fast and responsive, no matter how much traffic you receive. This requires a solid understanding of the underlying technologies running on your server, and how this impacts how quickly pages load. Let’s delve into HTTP, the protocol in which web servers and browsers “speak” to send and receive web page content.
HTTP is ultimately a request/response protocol. This means that you send a request to an HTTP server (e.g. Apache, IIS or Nginx), and the HTTP server generates a response and sends it back to the user.
Behind the scenes, Apache may pass the request to other processes (e.g. PHP-FPM) which execute code, run database queries, and generate the response which Apache then sends back to the client. This process is what takes up the bulk of “HTTP response time” that you’ll see in a performance benchmark (which will run through in more detail below).
We’ll provide some guidance on how to test site performance, what makes a site fast (or slow) and what we do to improve server response times.
Introducing TTFB: Time To First Byte
This response time is more accurately known as Time-To-First-Byte: i.e. how long it takes the web server to serve the first byte of an HTTP response to any given request. Our efforts in performance optimisation revolve around making sure this value is as low as possible.
So what is a good TTFB? This is difficult to give an absolute figure on because it varies depending on what you’re serving. Web servers are very fast at serving static resources- files such as images or CSS, which don’t need any extra computation to serve them, so can be delivered quickly. For reference, Google sees anything around 250ms or below as “good”. To put that in context, it takes humans around 400ms to blink!
Benchmark Performance testing
For extensive performance testing on your websites, we use our benchmark site performance tool – ProStack Webpage Test. This is a private instance of the webpagetest.org tool which provides results to demonstrate how your sites are performing, It’s also useful to quickly spot problems or get a feel for how fast a site is.
We use this test as an objective measure of how quickly a site is performing. Often we’ll run the test several times, before and after optimisation, to see how a specific change impacts site performance. As we mentioned above, TTFB is the key value we’re paying attention to here.
How we optimise your servers from behind the scenes
We’ll run through the key areas that we focus on to provide you with lightning-fast performance:
Caching – Many frameworks have built-in support for in-memory caching such as Redis or Memcache, or even an FPC (Full-Page-Cache) such as Varnish, so that frequently accessed resources don’t tax your web server or database. We can help implement these to improve overall Time To First Byte and page performance.
Database optimisations – We examine database performance and make sure things are optimal- in areas such as query caching… including query caching, buffer sizes and connection limits for busy sites.
Tuning – We can make sure your web server (be it Apache, Nginx or something else) can handle the volume of requests you’re expecting.
OS Optimisation – We have a suite of tried-and-tested tricks up our sleeve to optimise your server’s underlying operating system for the traffic your site will be receiving.
Performance optimisation can be part art, part science- every site is different and so we look at every layer of the stack to see how performance can be improved by specific changes.
My site is slow! How do I make it faster?
If you’ve noticed a slowness to your site and it doesn’t load as quickly as it used to, don’t panic!
Get in touch with our support team and we’ll look into it straight away, running through the steps provided above and advising on what we can do to get the most performance out of your server. We don’t settle for anything less, so we’ll make sure your site is running as fast as it should be.