Tutorials, best practices and news on Django, Python and modern web development.
A fast site does not stay fast on its own. Images, marketing scripts, database growth and Google's own thresholds all move. Here is what changes, the eight checks we run every month, and the database and image housekeeping that keeps a site quick.
TTFB is the wait between asking for a page and the first byte coming back: DNS, connection, redirects and the server's own work. Google calls 0.8 seconds or under good. Why it caps LCP, how it affects crawling, and where the time goes on a Django site.
The PageSpeed score is a weighted lab test, and four things drag most sites down: image weight, render-blocking CSS and JavaScript, third-party scripts and a slow server response. Here is the seven-day order we fix them in, and what each day should clear.
Google's Core Web Vitals are three field measurements: LCP (2.5 s), INP (200 ms) and CLS (0.1), judged at the 75th percentile of real visits over 28 days. What each one measures in the browser, how to read the two reports, and what usually fails.
Lazy loading tells the browser not to download an image until the visitor is about to scroll to it. One attribute does it. Here is how the browser decides, where it helps, the two mistakes that make a page slower, and the Django template we use.
We encoded real portfolio images to JPEG, WebP, AVIF and JPEG XL and compared them at equal visual quality. AVIF was 37% to 67% smaller than JPEG, WebP struggled on photos, and JPEG XL led at high quality but still needs a fallback in most browsers.
Largest Contentful Paint is the time from the click until the biggest image or text block on screen has rendered, and Google wants it at 2.5 seconds or under for three quarters of visits. Which element counts, where the time goes, and how to fix each part.
Cache versus CDN is not a choice. A cache is any stored copy of something expensive to fetch or build; a CDN is a network of servers that holds one near your visitors. How the browser, CDN and server layers work, what the headers mean, and what to cache where.