In memory computing principles by Mac Moore of GridGain

2. © 2014 GridGain Systems, Inc. Agenda • Overview • Why In Memory & Use Cases • Evolution of Architectures • Concepts and Considerations • In-‐Memory Data Fabric 6.5 • Data Grid • Clustering and Compute • Streaming • Hadoop Acceleration • Highlights: Release 6.5

3. © 2014 GridGain Systems, Inc. Why In-‐Memory Computing? Cloud/SaaS apps, Mobile Computing back-‐ends, Internet of Things, Big Data analytics, Social Networks – all need to be done in-‐memory to reach Internet scale “RAM is the new disk, disk is the new tape.” RAM is 3,000 times faster than spinning disks. By moving data from disk to RAM and employing modern in-‐memory data grid technology, things get fast. Really, really fast.

4. “In-‐memory computing will have a long term, disruptive impact by radically changing users’ expectations, application design principles, products’ architectures and vendors’ strategies.” In-‐memory computing is the future of computing… it offers a massive potential not only in TCO reduction but across all four value dimensions: performance, process innovation, simplification and flexibility. © 2014 GridGain Systems, Inc.

5. “Organizations that do not consider adopting in-‐memory application infrastructure technologies risk being out-‐ innovated by competitors that are early mainstream users of these capabilities” © 2014 GridGain Systems, Inc.

6. © 2014 GridGain Systems, Inc. In-‐Memory Computing: Why Now? In-memory will have an industry impact comparable to web and cloud. RAM is the new disk, and disk is the new tape.! In-Memory Computing Market:! • $13.23B in 2018! • 2013-2018 CAGR 43%! DRAM Cost, $ Cost Less than 2 zetabytes in 2011, 8 in 2015 drops 30% every 12 months BigData Technologies Planned 34% will use in-memory technology Data Growth

9. © 2014 GridGain Systems, Inc. In-‐Memory Technology: Use Cases Data Velocity, Data Volume, Real-‐Time Performance > Automated Trading Systems Real time analysis of trading positions & market risk. High volume transactions, ultra low latencies.! > Financial Services Fraud Detection, Risk Analysis, Insurance rating and modeling.! > Online & Mobile Advertising Real time decisions, geo-targeting & retail traffic information.! > Big Data Analytics Customer 360 view, real-time analysis of KPIs, up-to-the-second operational BI.! > Online Gaming Real-time back-ends for mobile and massively parallel games.! > Bioinformatics & Sciences High performance genome data matching, Environmental simulation.

11. © 2014 GridGain Systems, Inc. Traditional Architecture App Server1 User App Data Requests Traditional RDBMS Server Disks Disks Disks Disks Disks Disks App Server2 User App App Server3 User App App Server4 User App App Server n User App Processing Happens Here Data Requests Data Requests Data Requests Relational Data Data is converted Data is shipped across the network for every request to objects (Marshaling) All data is on disk. This is the central bottleneck. Beyond a certain point, scaling the RDBMS becomes complex, expensive and difficult to manage.

12. © 2014 GridGain Systems, Inc. Horizontally Scale the RDBMS Data is converted Scaling improves, but little else changes. Also these are very expensive! Data Requests RDBMS Server Server1 Server2 Disks Disks Disks Disks Disks Disks App Server1 User App App Server2 User App App Server3 User App App Server4 User App App Server n User App Processing Happens Here Data Requests Data Requests Data Requests Relational Data Data is shipped across the network for every request to objects (Marshaling) All data is still on disk.

13. © 2014 GridGain Systems, Inc. IMDG: Distributed Caching Scaling improves, as some (or all) data is now in RAM. But, we still have to ship data to the app for every request. Distributed Cache Server1 Server2 App Server1 User App App Server2 User App Traditional RDBMS Server Disks Disks Disks Disks Disks Disks App Server3 User App App Server4 User App App Server n User App Object Data RAM RAM RAM RAM RAM RAM Processing Happens Here Data is shipped across the network for every request

14. © 2014 GridGain Systems, Inc. GridGain: IMDG + Grid Compute By bringing computation to the In-‐Memory Data Grid, you can now achieve the fastest possible results. User App User App User App User App In-Memory Compute+Data Grid Processing is local, in-memory, and in native object format. All data is in RAM Server1 User App GridGain Node RAM RAM Server2 GridGain Node RAM RAM Server3 GridGain Node RAM RAM Server n User App GridGain Node RAM RAM Server4 User App GridGain Node RAM RAM Server5 GridGain Node RAM RAM Server6 GridGain Node RAM RAM Server n User App GridGain Node RAM RAM Tasks/Queries

15. © 2014 GridGain Systems, Inc. In-‐Memory Data Grids: Considerations Performance Scaling Consistency Search/Query Persistence Transactions Icons by: http://www.icons-‐land.com/ http://www.awicons.com/ http://www.aha-‐soft.com/

17. © 2014 GridGain Systems, Inc. In-‐Memory Data Grids: Considerations > Horizontal Scaling (“add a brick”) > Dynamic Topology (no downtime) > Vertical Scaling (large RAM allocation per-‐ process) > Good monitoring tools (utilization & mgmt.) Scaling

18. © 2014 GridGain Systems, Inc. In-‐Memory Data Grids: Considerations > Weak (Eventual) Consistency vs Strong Consistency > Per-‐store or per-‐cache configuration > Understand impact on performance Consistency

19. © 2014 GridGain Systems, Inc. In-‐Memory Data Grids: Considerations > Flexible Persistence Options > Read-‐Through, Write-‐Through, Write-‐ Behind > Examples: Relational Database, Local Storage, Shared Storage Persistence

20. © 2014 GridGain Systems, Inc. In-‐Memory Data Grids: Considerations > Are transactions supported? > Local or Distributed > Global or per-‐store/per cache configuration > Impact on performance Transactions

23. Data Velocity, Data Volume, Real-‐Time Performance © 2014 GridGain Systems, Inc. Customer Use Cases > Automated Trading Systems Real time analysis of trading positions & market risk. High volume transactions, ultra low latencies.! > Financial Services Fraud Detection, Risk Analysis, Insurance rating and modeling.! > Online & Mobile Advertising Real time decisions, geo-targeting & retail traffic information.! > Big Data Analytics Customer 360 view, real-time analysis of KPIs, up-to-the-second operational BI.! > Online Gaming Real-time back-ends for mobile and massively parallel games.! > Bioinformatics & Sciences High performance genome data matching, Environmental simulation.

24. Gartner names GridGain a 2014 “Cool Vendor” for IMC © 2014 GridGain Systems, Inc. GridGain named a 2014 AlwaysOn Global 250 Top Private Company “This positions GridGain as one of the few IMC open-‐source technologies available and the only one… providing such a rich set of functionality.”

25. Use Case: • Open © 2014 GridGain Systems, Inc. Largest bank in Eastern Europe (Russia), and the third largest in Europe tender won by GridGain –Goal: Real-‐time risk and leverage reporting on their global financial trading portfolio – Performed a detailed evaluation and software assurance test –Delivered best performance, scale and high availability 1 Billion Transactions per Second 10 Dell R610 servers 1 TB Memory < $40K

27. © 2014 GridGain Systems, Inc. GridGain In-‐Memory Data Fabric: Strategic Approach to IMC • Supports Applications of various types and languages • Open Source – Apache 2.0! • Simple Java/C#/C++ APIs! • 1 JAR Dependency! • High Performance & Scale! • Automatic Fault Tolerance! • Management/Monitoring! • Runs on Commodity Hardware • Supports existing & new data sources! • No need to rip & replace

28. • Distributed © 2014 GridGain Systems, Inc. In-‐Memory Key-‐ Value Store • Local, Replicated, Partitioned • TBs of data, of any type • On-‐Heap and Off-‐Heap Storage • Backup Replicas / Automatic Failover • Distributed ACID Transactions • SQL queries and JDBC driver • Collocation of Compute and Data In-‐Memory Data Grid

29. • Direct © 2014 GridGain Systems, Inc. API for MapReduce • Zero Clustering & Compute Deployment • Cron-‐like Task Scheduling • State Checkpoints • Early and Late Load Balancing • Automatic Failover • Full Cluster Management • Pluggable SPI Design

31. © 2014 GridGain Systems, Inc. Distributed Java Structures • Distributed Map (cache) • Distributed Set • Distributed Queue • CountDownLatch • AtomicLong • AtomicSequence • AtomicReference • Distributed ExecutorService

33. • Plug © 2014 GridGain Systems, Inc. Hadoop Accelerator and Play installation • 10x to 100x Acceleration • In-‐Memory Native MapReduce • In-‐Process Data Colocation • GGFS In-‐Memory File System • Pure In-‐Memory • Read-‐Through from HDFS • Write-‐Through to HDFS • Sync and Async Persistence

34. © 2014 GridGain Systems, Inc. In-‐Memory Accelerated Map Reduce • In-‐Memory Native Performance • Zero Code Change • Use existing MR code • Use existing Hive queries • No Name Node • No Network Noise • In-‐Process Data Colocation

37. © 2014 GridGain Systems, Inc. Cross-‐Language Interoperability • Portable Objects Language-‐neutral storage format. Allows you to write data from one language, and access or modify it from another. • Performance Across Languages Optimized neutral format that provides great performance across all supported languages. • Client Feature Parity Feature parity across supported language APIs. C# C++ Cross Language Data Interoperability Java GridGain Data Grid

38. © 2014 GridGain Systems, Inc. Dynamic Schema Changes • Dynamic Schema Changes Change data structures: add properties/fields dynamically at runtime when needed. • Searchable / Indexable Index and search into arbitrary fields, without needing to create annotated classes. • Version Independent Data storage format no longer tied to class versioning or product versioning. C# C++ Dynamic Schema Changes Language Independent Format Java

39. © 2014 GridGain Systems, Inc. Grid Managed Services • Automatic High-‐Availability Define services that should be instantiated by the grid. • Configurable Support for Grid Singletons, Node Singletons, and more. • Maintenance-‐free Deployment of services in the desired configuration is guaranteed by the GridGain Grid. Grid Managed Services Grid Singleton Node Singleton GridGain Data Grid GridGain Data Grid

40. © 2014 GridGain Systems, Inc. Enterprise Edition: Exclusive Features > Management & Monitoring GridGain Visor for GUI-based DevOps! > Local Restartable Store fast recovery during planned outages or DR! > Data Center Replication Multi-datacenter WAN support! > Network Segmentation Protection Configurable fault-tolerance for network interruptions! > Security Features Client Authentication and related SPIs! > Rolling Production Updates Perform software upgrades without downtime! > Support & Maintenance Support incidents, ticket access, upgrades, patches! > Training & Consulting Technical training and customized consulting services! > Deploy with Confidence Indemnification for Enterprise Customers

41. © 2014 GridGain Systems, Inc. Enterprise & Open Source Comparison Chart GridGain Enterprise Subscriptions include the following during the subscription term: > Right to Use the Enterprise Edition of the GridGain product. > Bug fixes, patches, updates and upgrades to the latest version of the product. > 9x5 or 24x7 Support for the product. > Ability to procure Training and Consulting Services from GridGain. > Confidence and protection, not provided under Open Source licensing, that only a commercial vendor can provide, such as Indemnification.

In memory computing principles by Mac Moore of GridGain

Subash DSouza

In memory computing principles by Mac Moore of GridGain