Some details
Chinese AdTech company experienced problems wanted to be prepared for the next level of market expansion. They wanted to be able to handle 10x traffic volume from a current peak load and it was their first goal. Because of a pay-per-click model, each second of downtime directly impact to their revenue. So making infrastructure highly available was a high-level goal for their business. The expertise of an in-house DevOps team
wasn',t enough to achieve both of these critical points and AdTech company decided to replace their in-house DevOps team with us.We came there with an audit of the current state and transfer of knowledge from the in-house DevOps team. They did not have any backup and monitoring procedures, so from this point data loss was excluded and the monitoring system started to collect performance and availability metrics. Then we discovered all single points of failure and implemented and tested clusterization and automatic failover solutions for them. We delivered aerospike, PostgreSQL and Redis clusters. The next point was the implementation of auto-scaling policy for Ad service nodes, so we automatically run temporary nodes during a high volume time frames and stop them during low volume time. As a result, we built an infrastructure which able to process 30 000 of requests per second and meet SLA requirements for 99.99% availability
