Varnish Memory & Storage Optimization

Varnish cache size grows with your catalog and traffic. If memory is undersized, cache objects get evicted frequently, lowering the hit ratio and increasing load on backend servers. Proper cache sizing keeps page delivery fast and stable during peak traffic.

1. Problem Statement: Varnish Showing Lower Hit Ratio

Varnish works best when it can retain enough cache objects in memory.
When cache storage is undersized:

Cache objects get evicted too frequently
Hit ratio drops
Backend servers receive more requests, increasing load and slowing page delivery

2. Diagnostics (Most Important Step)

A quick way to review important Varnish cache indicators is to check key counters from varnishstat.
varnishstat -1 | egrep “cache_hit|cache_miss|cache_hitpass|n_lru_nuked|n_object”
This command surfaces the core metrics needed to evaluate cache efficiency and memory pressure during Varnish cache optimization.

Cache Hit

Number of requests served directly from Varnish cache. Higher values indicate faster responses and lower backend load.

Cache Miss

Requests not found in cache and fetched from the backend. High miss rates can indicate weak caching or frequent cache invalidations.

Cache Hitpass

Requests intentionally configured to bypass caching. If this is high, many pages are not benefiting from Varnish caching.

n_lru_nuked (Evictions)

Objects removed from cache because storage was full. Frequent increases indicate that cache memory is undersized.

n_object

Total number of objects currently stored in cache. Helps understand how much content Varnish is holding.

TTL & Cache Fragmentation

Also review cache TTL values and cache tag invalidations. Very short TTLs or frequent purges can fragment the cache and reduce hit ratio even when sufficient memory is available for Varnish cache optimization.

3. Insights

In many ecommerce stores, Varnish storage is configured once during launch and rarely revisited. As the catalog grows and traffic increases, cache storage often becomes undersized, leading to more evictions and lower hit ratios affecting Varnish cache optimization.

4. Actions

Allocate Cache Memory

Allocate cache memory based on catalog size and traffic patterns. Adequate storage helps Varnish retain more objects and reduces frequent evictions.

Tune TTL and Grace

Set appropriate TTL and grace values so cached pages stay usable longer. Grace allows slightly stale content to be served while fresh content is fetched from the backend.