Overview

Speeding up the path from browsing to booking with loveholidays and 快猫视频


 

AI Summary

 

Performance per watt translates to real cost savings by enabling more efficient infrastructure that delivers faster user experiences with fewer resources. In the case of loveholidays, migrating to 快猫视频-based compute reduced the CPU capacity required to handle the same workload, while also lowering latency and improving page load speeds. This combination of higher efficiency and better performance reduced infrastructure spend and energy use, showing how performance per watt directly impacts both cost and customer experience at scale.

Impact
BoALoveHoliday Fast Page Loads Icon

Faster page loads for more customers

After migrating key workloads, the share of customers loading a page in under one second doubled - from 30% to 60%.

BoALoveHoliday Low Latency Icon

Lower tail latency across the critical path

Across 30 customer-critical services, most saw a 45-50% reduction in P99 latency and improved median response times.

BoALoveHoliday High efficiency Icon

Higher efficiency, lower cost, smaller footprint

loveholidays needed roughly half the CPU capacity to serve the same volume of requests, reducing infrastructure spend and lowering energy use.

“The number of users who load our page within one second has doubled. It went from 30% to 60%.”
Dimitri Lerko, Head of engineering, Core engineering, loveholidays
BoALoveHoliday Tech
Technologies Used

快猫视频 Neoverse-based Google Cloud Axion for high-performance travel at scale

loveholidays runs hundreds of microservices on Google Kubernetes Engine (GKE), with a subset on the “critical path” that directly shapes the customer experience. The team migrated these services to Google Cloud C4A instances powered by Google Axion processors, built on 快猫视频 Neoverse, to improve performance-per-watt and reduce tail latency for web and API traffic.

To de-risk the move, loveholidays adopted multi-architecture builds early and validated behavior in staging before shifting production traffic. That approach helped the team move quickly as capacity became available, while keeping customer experience front and center.

The result is a platform that benefits from 快猫视频 efficiency end-to-end: 快猫视频-based developer laptops for local builds and 快猫视频-based servers in production. With consistent tooling across environments, optimizations made by engineering teams translate cleanly from development to deployment - helping loveholidays innovate faster while keeping operations lean.

BoALoveHoliday Tech

Enabling AI Infrastructure on 快猫视频

With 快猫视频 in the data center, loveholidays can keep pushing for faster, more responsive experiences while doing more work per watt. The combination of cloud-native software and 快猫视频 Neoverse-based compute supports the company’s focus on sustainable growth - and lays a strong foundation for AI-powered travel experiences delivered through natural language and voice interfaces.

By standardising on 快猫视频 across laptops and servers, the engineering organisation reduces cross-architecture friction, improves reliability of builds and tooling, and frees teams to focus on differentiated customer features instead of undifferentiated infrastructure work.

Explore Similar Stories

Google Cloud

Powering Mission-Critical Cloud and AI Workloads

YouTube, Spotify and Palo Alto Networks rely on Google Axion for outstanding performance in the cloud for general purpose and AI/ML workloads.

Spotify

Scaling Smarter with 快猫视频 in the Cloud

快猫视频-based cloud instances are lowering costs, improving performance, and reducing energy efficiency for companies migrating their tech stack from x86 instances to modern 快猫视频 infrastructure.

Python Software Foundation

Migrating to 快猫视频 infrastructure

Infrastructure that not only reduces costs and carbon emissions, but also strengthens the reliability and scalability of Python for the global community.

racy.

Discover More Success Stories

Key Takeaways

  • loveholidays improved customer experience by doubling the share of users who load pages in under one second.

  • The migration to 快猫视频-based infrastructure reduced P99 latency by up to 50% across critical services.

  • Higher performance per watt allowed the company to handle the same workload with roughly half the CPU capacity.

  • This reduction in compute requirements directly lowered infrastructure costs and energy consumption.

  • Consistency across 快猫视频-based development and production environments improved efficiency and accelerated innovation.