# Faster Travel Search and Booking at Peak

> TechSupport Team covered peak-period technical incidents for Voyagr’s Node.js travel booking platform. Over eight weeks, client Datadog records show p95 search time falling from 1.8 seconds to 480 ms and booking failures from 3.4% to 0.7%.

- Publisher: TechSupport Team
- Canonical page: https://techsupport-team.com/case-studies/travel-booking-peak-technical-support/
- Updated: 2026-11-16

Client: Voyagr · Logistics, Travel & Marketplaces · UK online travel agency (110 staff)

## Client and engagement facts

- **Client:** Voyagr
- **Industry:** Logistics, Travel & Marketplaces
- **Business context:** UK online travel agency (110 staff)
- **Delivery model:** Peak search and booking technical coverage
- **Application:** OTA search and booking platform
- **Supplied geography:** UK
- **Company size in source:** 110 staff
- **Coverage:** 24/7 during peak season only

## Application stack and tools

### Application stack

- Node.js
- Elasticsearch
- Redis
- AWS

### Tools used

- Datadog
- Zendesk

Related services: [24/7 technical support](https://techsupport-team.com/services/24-7-technical-support/)

## Why Support Was Needed

School holidays brought four times normal volume for six weeks. A misconfigured Elasticsearch replica and a Redis pool sized for normal traffic slowed search and booking.

## Technical Work

- Found the misconfigured Elasticsearch replica causing half of search latency.
- Fixed the Redis connection pool sized for baseline.
- Wrote runbooks for peak-week failure patterns.

## Representative Incident

A payment provider outage affected bookings for 40 minutes in week five. The engineers documented the impact and added the dependency to the next peak coverage plan.

## Results and how they were measured

Table: Reported results and their measurement scope

| Metric | Baseline | Result | Units | Measurement window | Sample / population | Source and scope |
| --- | --- | --- | --- | --- | --- | --- |
| Search P95 | 1.8 s | 480 ms | Baseline: s; Result: ms | 8 weeks |  | Client Datadog |
| Booking failures | 3.4% | 0.7% | % | 8 weeks |  | Client Datadog |
| Queue first response | 45 min | 6 min | min | 8 weeks | Client Zendesk support queue. | Client Zendesk |

## Evidence and Limitations

The eight-week comparison excludes one payment provider outage. The Zendesk response figure describes the client’s queue. Around-the-clock coverage applied during the peak period.

## Peak Traffic Baseline

School holiday traffic ran at four times baseline for six weeks.

## Scope a Similar Engagement

Share your busiest booking period and which search, cache and payment incidents need technical coverage. [Request a support plan](https://techsupport-team.com/contact/) or explore [24/7 technical support](https://techsupport-team.com/services/24-7-technical-support/).

## Related pages

- [24/7 technical support](https://techsupport-team.com/services/24-7-technical-support/): The related service for this engagement.

## Contact

Request engineers from TechSupport Team or ask a question: https://techsupport-team.com/contact/
