Admiral Scales Globally with YugabyteDB on GCP + Single-Digit Latency

on YouTube ↗23m

" summary

Admiral migrates from NoSQL to YugabyteDB on GCP to serve publisher account data with global single-digit reads, running a Go microservice stack across 5 regions, 3 continents with two instances per region, RF=3 and three US masters plus regional read replicas. Deployment uses Ansible on SSD persistent disks with TLS and encryption at rest, a Go CQL driver patch to prefer one replica, and achieves 50th <3ms, 90th = 5ms reads at 4000 QPS peak while 90th writes are ~230ms.

AI-written summary · about

" chapters (10)

1. 0:00Start
2. 0:12What does Admiral do?
3. 2:12Publishers are building first-party relationships
4. 4:00Admiral Accounts
5. 5:38Our Tech Stack
6. 6:39Requirements
7. 9:03Previous Attempts
8. 12:57Solution: YugabyteDB
9. 16:24Implementation
10. 18:48Performance

" related talks

• 19mHow Admiral Scales YugabyteDB Effortlessly on Google CloudJames Hartig · Distributed SQL Summit Asia 2022
• 30mFrom monolith to petabyte scale: Built a cloud-native, AI-ready data layerSanjeev Gautam · Distributed SQL Summit Mumbai 2026
• 26mA Winning Multi Cloud Data StrategyChirag Narang · Distributed SQL Summit Asia 2021
• 29mEnterprise Scale: How a Distributed Architecture Can Deliver Scale and ResilienceAmit Chauhan, Rahul Batta · Distributed SQL Summit 2022
• 57mNo Downtime, No LimitsKarthik Rangathan · Distributed SQL Summit Asia 2024
• 30mBuilding Cloud Native Applications on Distributed SQL PlatformSHIGERU URUSHIBARA · Distributed SQL Summit Asia 2023
• 30mTechnical Comparison of YugabyteDB with Oracle and PostgreSQLFranck Pachot · Distributed SQL Summit Asia 2024
• 25mDiscover the Easiest Path to Distributed DBaaS: Yugabyte CloudAmol Bhoite · Distributed SQL Summit Asia 2022