Speed Caching: Reducing Supplier Load by 80%

What is Speed Caching?

Travelgate's Speed Cache layer sits between your Search queries and the upstream supplier connections. Instead of querying each supplier in real time on every search, the cache serves pre-fetched, periodically refreshed results — dramatically reducing response times and supplier API call volumes.

At scale, a well-configured cache reduces supplier load by 75–85% while delivering sub-200ms search responses to end users.

Standard vs Fast Mode

Standard mode fetches and caches results on a scheduled refresh cycle. Fast mode uses a background pre-fetch triggered by the first cache miss: the first user to search a destination/date combination gets a live response while the cache is populated for subsequent users.

Fast mode is recommended for long-tail destination searches where pre-seeding the cache is impractical. Standard mode is best for high-volume routes where cache warm-up can be managed proactively.

TTL Tuning Per Booking Window

Time-to-live (TTL) should vary by how far in advance the search is occurring. A search for travel 60+ days out can tolerate a longer TTL (4–8 hours) because price volatility is lower. A search for travel within 7 days should use a shorter TTL (15–30 minutes) to reduce the risk of serving stale availability.

Configure TTL rules in the Travelgate Speed Cache settings per supplier and per booking-window bucket. Review and adjust quarterly as your traffic patterns evolve.

Monitoring Cache Hit Rates

A healthy cache hit rate for a mature integration is 70–85%. Rates below 50% indicate either an under-seeded cache (increase pre-fetch frequency), a TTL that is too short (increase for advance-booking windows), or extremely long-tail searches with no repeat demand.

Travelgate's Stats module exposes hit/miss metrics per supplier and per destination. Set alerts for hit rates below 60% and investigate root causes before they impact search performance.

The Critical Relationship Between TTL and DeltaPrice

Every cached Search result will be validated by a live Quote call before booking. The longer your TTL, the greater the probability that the quoted price differs from the cached search price — triggering a DeltaPrice event.

Calibrate your DeltaPrice threshold to your TTL. A 4-hour TTL might produce a 3–5% price drift on average; set your auto-accept threshold at 5% and your hard-reject threshold at 15%. Review these figures monthly against your actual DeltaPrice distribution.

Try It Yourself