Skip to content
Tutorials Center
My learning

Topics

Lesson 34 of 37 ยท Deployment and Performance

Performance Tuning and Updates

Make Drupal fast and keep it healthy: PHP and database tuning, aggregation, image optimization, Redis, CDN, queues, cron, Views caching, monitoring and a measuring stand-in.

Drupal can serve millions of page views, but only when it is configured for production and kept healthy. Speed comes from layers: efficient PHP and database, good caching (caching and cache tags), light pages and a CDN. Health comes from cron, updates, logs and monitoring. This lesson gives you a practical checklist, ordered by effect.

Drupal cannot run on this site, so configuration and commands are read-only. The TryIt blocks measure and reason about performance ideas with JavaScript stand-ins.

1. Measure first

Do not tune blindly. Use:

  • Browser tools (Lighthouse, network panel) for page weight and Core Web Vitals.
  • Response headers with cacheability debugging to see cache hits.
  • Web Profiler / Devel on development sites for queries per request and slow components. Remove them in production.
  • Slow query log in MySQL (long_query_time = 1) and EXPLAIN for the worst queries.
  • APM tools such as Blackfire, New Relic or Tideways for real PHP profiles.
Node.js Stand-in: percentiles beat averages

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with node:readline or process.stdin), it gets the next line from this box.

2. Production settings

  • Performance page (/admin/config/development/performance): enable page cache, set a browser/proxy max-age (for example 1 hour to 1 day), and turn on CSS and JavaScript aggregation.
  • Install Internal Dynamic Page Cache and Internal Page Cache (both are standard) and, for logged-in traffic, BigPipe.
  • Use PHP 8.3+ with OPcache (opcache.memory_consumption=256, opcache.validate_timestamps=0 with a reset on deploy) and enough memory_limit (256M+).
  • twig.config: debug: false, auto_reload: false and cache: true in production.
  • Disable dev modules (Devel, Kint, Webprofiler, Field UI and Views UI if not needed).

3. Database and Redis

Keep MySQL/MariaDB or PostgreSQL tuned: enough InnoDB buffer pool (innodb_buffer_pool_size about 50 to 70 percent of RAM on a dedicated database server), indexes on columns you filter and sort by, and a connection pool sized for your PHP workers.

Move caches out of the database with Redis (module redis) or Memcache:

PHP
$settings['redis.connection']['interface'] = 'PhpRedis';
$settings['redis.connection']['host'] = getenv('REDIS_HOST') ?: '127.0.0.1';
$settings['cache']['default'] = 'cache.backend.redis';
$settings['cache']['bins']['bootstrap'] = 'cache.backend.chainedfast';

This removes thousands of small reads and writes from the database and speeds up every request, especially cache rebuilds.

4. Views and queries

  • Enable Views caching (tag-based) on blocks and pages, and use pagers instead of "all items".
  • Avoid fetching many fields or relationships you do not display; add exposed filters on indexed columns.
  • Do not load full entities in loops (the N+1 problem). Use loadMultiple() and render with view modes, or use Views fields rather than rendered entities.
  • Prefer entity queries to raw SQL, and add indexes with hook_schema or a hook_update_N for custom tables.
Node.js Stand-in: the N+1 query problem

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with node:readline or process.stdin), it gets the next line from this box.

5. Make pages light

  • Images: use image styles and responsive images with srcset, serve WebP/AVIF (the core image toolkit can convert with the webp image style effect), set loading="lazy" (on by default for image fields) and always provide width and height.
  • JavaScript and CSS: attach libraries only where needed (preprocess and libraries), avoid jQuery-only dependencies in new code, and defer third-party scripts.
  • Fonts: self-host, subset and font-display: swap.
  • Core Web Vitals: reserve space for images and embeds to avoid layout shift; keep the largest above-the-fold element simple.

6. CDN and reverse proxy

Put a CDN (Cloudflare, Fastly, CloudFront) or Varnish in front of anonymous traffic. Drupal sends Cache-Control: public, max-age=... and X-Drupal-Cache-Tags, so modules such as Purge can invalidate the edge when content changes. Serve /sites/default/files and static assets from the CDN. Make sure personalised pages send Cache-Control: private (Drupal does this for logged-in users) and that cookies on anonymous pages do not defeat caching.

7. Cron, queues and background work

Cron runs housekeeping: clearing caches, indexing for search, sending queued mail, scheduled publishing and any module's hook_cron. Do not rely on Automated Cron on a busy site; trigger it from the system:

Terminal
# crontab -e
*/10 * * * * cd /var/www/site && vendor/bin/drush cron

For heavy tasks (sending 10,000 emails, importing files), put work in a queue: a Queue Worker plugin processes items in small batches during cron or with drush queue:run my_queue. This keeps web requests quick and gives retries and rate limits. The Batch API does the same for interactive operations with a progress bar.

Node.js Stand-in: a queue worker with time limit

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with node:readline or process.stdin), it gets the next line from this box.

8. Keep the site healthy

  • Updates: subscribe to security advisories and apply core and module updates with Composer on a schedule; test on staging (deployment). Use drush pm:security and the Status report.
  • Logging: send logs to syslog or an external service instead of the database (dblog grows). Set drush watchdog:delete all regularly if you must keep it.
  • Monitoring: uptime checks, error-rate alerts, disk space, queue length and cron success.
  • Search: use Search API with Solr or Elasticsearch for large sites instead of core search.
  • Backups: automatic, off-server, restore-tested.
Terminal
drush status
drush core:requirements
drush watchdog:show --count=20 --severity=Error
drush sql:dump --gzip --result-file=../backups/site-$(date +%F).sql

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What is the first step in performance work?

    Measurements show where time is lost, so you fix the right thing.
  2. 2. Why move cache bins to Redis?

    A fast in-memory store cuts database load for every request.
  3. 3. What is the N+1 problem?

    loadMultiple and joins avoid the many repeated queries.
  4. 4. Where should heavy background work go?

    Queues keep requests fast and allow retries and batching.

Summary

  • Measure with percentiles and profilers before tuning.
  • Enable aggregation, page caches, OPcache and Redis in production; turn Twig debug and dev modules off.
  • Avoid N+1 queries, cache Views, serve light and responsive assets, and use a CDN with cache tags.
  • Run cron from the system and move slow jobs into queues.
  • Keep the site healthy with updates, log management, monitoring and tested backups.

Next: Project: Event Listing Site