MULTI-DIMENSIONAL SCALING – A NEW ARCHITECTURE FOR SCALING BIG DATA APPLICATION NoSQL Night Singapore Clarence J M Tauro Sr. Instructor Couchbase 

©2015 Couchbase Inc. 2 Agenda  Brief history of scaling in database  Scaling up  Scaling out  NoSQL workloads and scalability model  Core data operations, indexing and querying  Homogenous HW scaling  Introducing multi-dimensional scaling  Isolation of workloads through a services architecture  Independent HW scaling  Q & A 

©2015 Couchbase Inc. 3 Disclaimer & Notes  Disclaimer: The views expressed in this presentation are our own and do not necessarily reflect the views of Couchbase  Most of the content in this presentation is originally created by Anil, Product Manager, Couchbase  Thanks Anil, for allowing me to re-use your slides  3 

©2015 Couchbase Inc. 4 About the Speaker  Clarence J M Tauro – [email protected] - Senior Instructor, Couchbase - ~11 Years Professional Teaching and Consulting Experience - Worked at Pivotal – Instructor/Consultant for Spring/Spring Security/Spring Web/Enterprise Integration with Spring/Spring JMS/Spring Web/Spring Batch, Pivotal Hadoop/Cloud Foundry - PhD in Computer Science from Christ University [thesis accepted] - Hard-core Dog lover 4 

Brief History of Scaling in Database 

©2015 Couchbase Inc. 6 Scaling up  Scale-up architecture  Cluster processors – hyper-threading to cores  Locally partition workload among processors  Communicate over memory 

©2015 Couchbase Inc. 7 Scaling Up – Pros & Cons Pros  Can result in major performance improvement  Machines can now support having many cores and terabytes of RAM Cons  Expensive  Requires downtime  Performance bounded – at some point the database engine itself becomes the bottleneck  Limited in scalability and elasticity 

©2015 Couchbase Inc. 8 Scaling out  Scale-out architecture  Cluster of commodity HW  Horizontal partitioning of data on cluster nodes  Communicate over network 

©2015 Couchbase Inc. 9 Scaling Out – Pros & Cons Pros  Simple and easily scalable  Data evenly split across cluster of nodes  Scales linearly with throughput  Highly available  No single point of failure Cons  Not great for all workloads – data, index and query 

©2015 Couchbase Inc. 10 So Which Model is the Right Model ? Scale up or scale out? 

©2015 Couchbase Inc. 11 Scaling Up vs. Scale Out Link to whitepaper - http://www.msr- waypoint.com/pubs/204499/ a20-appuswamy.pdf 

NoSQL Workloads & Scalability Model 

©2015 Couchbase Inc. 13 NoSQL Workloads  One database, many workloads  Core data processing: GETs & SETs for a given key  Indexing: Index maintenance and lookups  Querying: Combine index and data with complex just-in-time data re-shaping, ordering, grouping, aggregations, and more Varying resource requirements - CPU, RAM, I/O, Network Varying methods to optimize latency & throughput for each 

©2015 Couchbase Inc. 14 Scalability Model Today Homogenous Scaling  Each node get a slice of the workload  Simple to do… But...  Workloads compete and interfere with each other  Can’t fine tune each workload  Core Data operation are partition-able so great with wider fan-out  Indexing and queries aren’t always partitionable, so worse with wider fan-out Index Workload Couchbase Cluster Query Workload Data Workload node1 node8 

Introducing Multi-Dimensional Scaling 

©2015 Couchbase Inc. 16 Modern Architecture What is Multi-Dimensional Scalability? MDS is the architecture that enables independent scaling of data, query and indexing workloads. Index Service Couchbase Cluster Query Service Data Service node1 node8 

©2015 Couchbase Inc. 17 Couchbase Cluster node1 node8 Modern Architecture  Isolated Service for minimized interference  Independent “zones” for Query, Index and Data Services Minimize indexing and query overhead on core key-value operations. Index Service Query Service Data Service 

©2015 Couchbase Inc. 18 Modern Architecture  Independent Scalability for Best Computational Capacity per Service Heavier indexing (index more fields) : scale up index service nodes More RAM for query processing: scale up query service nodes Couchbase Cluster node1 node8 node9 Data Service Index Service Query Service 

Services Architecture Data, Index, & Query 

©2015 Couchbase Inc. 20 Full Cluster Architecture STORAGE Couchbase Server 1 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 2 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 3 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 4 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 5 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 6 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service 

©2015 Couchbase Inc. 21 Full Cluster Architecture STORAGE Couchbase Server 1 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 2 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 3 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 4 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 5 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service STORAGE Couchbase Server 6 SHARD 7 SHARD 9 SHARD 5 SHARD SHARD SHARD Managed Cache Cluster Manager Cluster Manager Managed Cache Storage Data Service Index Service Query Service 

©2015 Couchbase Inc. 22 Storage Engine Cluster Manager Data Service Projector & Router New Services in Couchbase Server 4.0 Query Service Index Service Supervisor Index maintenance & Scan coordinator Index#2 Index#1 Query Processor cbq-engine Bucket#1 Bucket#2 DCP Stream Index#4 Index#3 ... B u c k e t # 2 B u c k e t # 1 18093 11211 18901 Managed Cache 

©2015 Couchbase Inc. 23 Recap  MDS enables unprecedented control of scalability with Couchbase Server  Separate out competing workloads to independent services  Independently scale each service “zone” within the cluster  Couchbase Server with MDS maximizes scalability and performance  Improves scale and performance to degrees not possible with other NoSQL or big-data engines on premise or in the cloud  Improved price/performance and squeezes more performance and throughput for mission-critical systems 

Thank you. 

Get Started with Couchbase Server 4.0: www.couchbase.com/beta Get Trained on Couchbase: http://training.couchbase.com CD220: Developing Couchbase NoSQL Applications Oct 20 – Oct 23 2015 CS300: Couchbase NoSQL Server Administration Nov 17 – Nov 20 Enroll Today!