FindHotel streamlines ECS workload automation while reducing costs with Elastigroup by Spot
Challenge – Automating and Optimizing Infra for Mission-Critical ECS Workloads
Launched in 2010 as a cloud-native company, FindHotel was already using many AWS services such as Elastic Container Service (ECS), S3, Firehose and even EC2 Spot instances for cost-reduction in some QA and Dev environments.
As a hotel search aggregator, they were using Amazon’s ECS to run their “Rates and Availability API”, a mission critical component in FindHotel’s travel platform.
As their EC2 costs for ECS rose with increased searches on their platform, FindHotel CTO, German Gomez-Herrero started to contemplate spot instances which offered up to 80% cost reduction.
German was considering building out their own logic to handle autoscaling, improve utilization, and leverage spot instances all while handling interruptions and maintaining high availability.
However, he was concerned that a DIY approach of developing and managing the underlying infrastructure for ECS would increase the overhead for his already busy DevOps team.
Additionally, without any SLAs for EC2 Spot instances expanding Spot usage to FindHotel’s core, high-availability applications running on ECS, was simply not an option.
Solution – Automated Infra Orchestration for Containers
After researching available solutions, Gomez-Herrero discovered Elastigroup by Spot for ECS. Elastigroup automatically handles the provisioning, management, and intelligent scaling of the underlying cluster while ensuring high availability at significantly lower costs.
With just a few clicks in the Spot dashboard, FindHotel was able to generate a Terraform template to migrate their workloads into Elastigroup by Spot for ECS.
German added that “we already were automating our AWS operations with Terraform, so the seamless integration with Elastigroup really shortened our time-to-value”.
After completing the initial migration, Elastigroup by Spot then began autoscaling preferred EC2 instance types to support the specific ECS Task and Service requirements for CPU, Memory and available Ports.
By optimizing task placement across the cluster (aka Tetris Scaling) and by automatically managing a buffer of spare CPU and Memory capacity, Elastigroup by Spot ensured optimal cluster utilization and availability without any need for human intervention and management.
On top of all this “Spot was able to run FindHotel’s mission-critical workload entirely on affordable spot instances with guaranteed uptime so our end-users could always search for the best hotel deals available” enthused Gomez-Herrero. He added “I would definitely recommend Spot to other DevOps practitioners as an easy-to-use and high-value cloud workload automation platform”.
Results and Benefits – Simplified Infra Management, Enterprise SLA and Major Cost Reduction
Some of the benefits that were realized by FindHotel include:
Quick Time-to-Value with Zero Risk and Minimal Effort
During the Spot POC, FindHotel’s engineers easily migrated their largest, high availability system from Fargate to Elastigroup in just a few short days and very quickly achieved massive cost reduction on their cloud-compute spend.
Advanced Cloud Workload Automation Designed for ECS
Even with fluctuating search traffic on the FindHotel website, Spot automatically managed the number and size of instances needed to support all Task requests. This intelligent automation freed up their DevOps team to focus on building applications, rather than managing the infrastructure.
70% Cost Reduction with High Availability SLA
After testing Spot’s SLA and running their clusters on a mix of Spot and On-Demand Instances, FindHotel gained the confidence to move their entire ECS workload to spot instances, helping reduce their cloud compute spend by 70% while being assured that their travel site would be available at all times.
ECS Cost Analysis by Services, Applications, and Tasks
FindHotel gained clarity into cost allocation using Elastigroup visualization of their EC2 spend broken down to the container level. This provided better governance and allowed them to ascertain ROI for business operations.
90% Resource Utilization
Elastigroup by Spot studied FindHotel’s actual container requirements and matched ECS Tasks and Services with the most efficient and optimized mix of EC2 Instances. This resulted in 90% instance utilization and additional cost optimization.