Introducing YugaByte DB.ppt

1、Introducing YugaByte DB,Karthik Ranganathan Co-Founder & CTO November 13, 2017,Mission,Simplify Mission-Critical Data,About Us,Kannan Muthukkaruppan, CEO Nutanix Facebook Oracle IIT-Madras, University of California-Berkeley,Karthik Ranganathan, CTO Nutanix Facebook MicrosoftIIT-Madras, University of

2、 Texas-Austin,Mikhail Bautin, Software Architect Clear Story Data Facebook D.E.Shaw Nizhny Novgorod State University, Stony Brook,Built Facebooks NoSQL platform powered by Apache HBaseScaled the platform to serve 10 mission-critical use cases Messages (OLTP) Operational Data Store (Time Series)Reass

3、embled the same Facebook team at YugaByte along with engineers from Oracle, Google, Nutanix and LinkedInHands-on with Apache HBase, Apache Cassandra, Oracle, RocksDB,Founders,Investors,Ravi Mhatre Founding Investor at Nutanix AppDynamics Mulesoft Lightspeed Venture Partners, and prominent angels,Jef

4、f Rothschild Former VP, Infrastructure Software at Facebook,What?,Converge operational database needs of mission-critical apps into a cloud-native platform,Cloud-Native, Mission-Critical, Open-Source,Breaking Barriers,A unified, cloud-native platform for mission-critical data,SQL DB Reliable, strong

5、ly consistent core Relational data model & ACID txns Low-latency reads & writes,NoSQL DB Available across multi-datacenters Multi-model High throughput writes, low-latency reads Native integration w/ Analytics & Streaming infrastructure (e.g. Apache Spark/Kafka),Cloud-Native Infrastructure Elastic a

6、nd geo-distributed fault-tolerance Reconfigurable with zero downtime Cloud-agnostic, portable across clouds,Cloud-Based Mission-Critical Apps Today,Datacenter 1,SQL Master,SQL Slave,Stateless microservices,Datacenter 2,SQL for OLTP data Manual sharding Cost: dev team,Manual replication Manual failov

7、er Cost: ops team,NoSQL for other data App aware of data silo Cost: dev team,Cache for low latency App does caching Cost: dev team,Data inconsistency/loss Fragile infra Hours of debugging Cost: dev + ops team,Cloud-Based Mission-Critical Apps with YugaByte,Datacenter 1,Stateless microservices,Datace

8、nter 2,Automatic Sharding Automatic Failover Cost: dev team,No data silos Cost: dev team,Transparent caching Cost: dev team,Build better apps Lower total cost High agility,Architecture 3 Node Cluster,DocDB LSM Storage Engine Built for storing ever-growing data, based on a heavily customized version

9、of RocksDB,tablet 1,tablet 1,yb-tserver1,yb-tserver2,yb-tserver3,tablet 1,YB-TServer Host & serve user data,Raft Consensus Replication Highly resilient, used for both data replication & leader election,Powering Mission-Critical Use Cases,tablet 1,tablet 1,yb-tserver1,yb-tserver2,yb-tserver3,tablet 1

10、,YB-Master Manage shard metadata & coordinate system-wide ops,YB-TServer Host & serve user data,Consistency Zero data loss persistence,yb-master1,yb-master3,yb-master2,Powering Mission-Critical Use Cases,tablet 1,tablet 1,yb-tserver1,yb-tserver2,yb-tserver3,tablet 1,YB-Master Manage shard metadata &

11、 coordinate system-wide ops,YB-TServer Host & serve user data,Availability Resilient under any failure,yb-master1,yb-master3,yb-master2,Powering Mission-Critical Use Cases,Performance No quorum reads 3x throughput of traditional quorum-based NoSQL,tablet 1,tablet 1,yb-tserver1,yb-tserver2,yb-tserver

12、3,tablet 1,YB-Master Manage shard metadata & coordinate system-wide ops,YB-TServer Host & serve user data,yb-master1,yb-master3,yb-master2, With Cloud-Native Agility,Reconfigure hardware | AZ | region |cloud with zero downtime,YB-Master Manage shard metadata & coordinate system-wide ops,YB-TServer H

13、ost & serve user data,Old infrastructure,New infrastructure,YugaByte Admin Console,Consistent CX,Lower TCO,Increase Agility,Business Value Proposition,1. MySQL master-slave replication,2. Cassandra cross-DC queue for cache updates,3. Per-DC Couchbase for caching,Current State,Case Study #1 - User Id

14、entity Login, change password, view profile,With YugaByte EE,One click deploy of sync and async replication,Case Study #1 - User Identity Login, change password, view profile,Unified platformZero data loss even on DC failuresAdd new DCs with ease,Redis cluster for low latency reads,Fragile (manually

15、 sharded & load balanced) Expensive (entire dataset in memory),On-premises only, would like to run hybrid/public cloud deployments,DB,Current State,Case Study #2 - Financial Data Service,With YugaByte EE,One click deploy of unified Redis cluster + DB,Case Study #2 - Financial Data Service,Develop mo

16、re business use casesOnly hot data in memory, auto-sharding and load balancingHybrid cloud deployment and ability to burst,Product editions,The simplest way to run YugaByte DB in production,Apache Cassandra & Redis APIs w/ Single Row ACID txns,TODAY,Scale-out, Transactional, Cloud-Native Data Fabric,SQL API w/ Distributed ACID txns,Graph, Full-text Search, Large Object Store ,ROADMAP,Simplify Mission-Critical Data,Demo,