cray urika-xa advanced analytics platform

10
Urika-XA Press Deck

Upload: insidehpc

Post on 29-Nov-2014

230 views

Category:

Technology


2 download

DESCRIPTION

In this slidecast, Ramesh Menon from Cray introduces the Urika-XA Advanced Analytics Platform. The Cray Urika-XA system provides customers with the benefits of a turnkey analytics appliance combined with a flexible, open platform that can be modified for future analytics workloads. http://cray.com Watch the video presentation: http://wp.me/p3RLHQ-dd7

TRANSCRIPT

Page 1: Cray Urika-XA Advanced Analytics Platform

Urika-XA Press Deck

Page 2: Cray Urika-XA Advanced Analytics Platform

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Turnkey Advanced Analytics

Platform

Next-Generation System Architecture

Engineered for Performance

The Urika-XA Advanced Analytics Platform

• Hadoop and Spark ecosystem • Emerging analytic workloads • Open platform for current and future frameworks • Single pane of glass for system management

• Innovative use of storage technologies • Battle-tested on cutting-edge

government/scientific analytic applications • Ready for the enterprise

• Dense footprint: over 1,500 cores, 6TB memory • 38TB SSD and 120TB POSIX-compliant high-

performance storage • InfiniBand • Cray Adaptive Runtime for Hadoop • Scale out to multi-rack configurations

Page 3: Cray Urika-XA Advanced Analytics Platform

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Urika-XA Single Rack Configuration • 48 Compute Nodes

• High-performance Intel processors • Infiniband • 800GB PCIe SSD per node

• Optimal combination of high-performance storage • 200TB total SSD, HDD and Lustre (Sonexion 900) storage • HDFS compatibility and POSIX compliance • Includes full Lustre HA capabilities

• Software stack • Cloudera Enterprise • Apache Spark • Cray Adaptive Runtime for Hadoop • Urika-XA Management System

• Multi-rack configurations available 3

Page 4: Cray Urika-XA Advanced Analytics Platform

Single, Multi-Use Analytics Platform Needed

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

ETL Stream Processing

Data Mining

Interactive Queries

Actionable Insight

Multiple steps of analytics processing • Batch, interactive, streaming • Low-latency applications require performance

optimizations

Analytics Pipeline

Cluster sprawl to handle variety of analytics • Large datacenter footprint • High management cost • Significant data movement • High TCO

Page 5: Cray Urika-XA Advanced Analytics Platform

Integrated, Open Platform Preferred

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Roll Your Own ✔ ︎Flexibility to support current

and future analytics workloads ✘ Complex to set up and manage ✘ Conventional big data

architecture not compliant with IT standards

✘ Hadoop stack not well integrated

Appliance ✔ Pre-integrated hardware and software ✘ Locked into vendor’s software stack ✘ Cannot update analytics platform as

big data analytics technologies evolve

Cloud ✔ Accelerated time to value ✘ Loss of control over data ✘ Data movement is expensive ✘ Lacks performance optimizations

Preferred Solution ✔ Pre-integrated hardware and software ✔ Accelerated time to value ✔ Flexibility to support current and future analytics

workloads

Page 6: Cray Urika-XA Advanced Analytics Platform

Convergence of Analytics and HPC

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Next-Generation Analytics Requires High-Performance Architectures

High-Performance Data Analysis • Finance: portfolio optimization, pricing, risk • Energy: seismic modeling • Life sciences: genomics, drug discovery • Scientific: simulation, weather forecasting

Traditional Big Data • standalone processing frameworks • batch analytics

Integrated Analytic Platform • versatile, multi-use • no data movement • low-latency, high-performance, and batch

“Simulation is the original Big Data Market” – IDC

Page 7: Cray Urika-XA Advanced Analytics Platform

Urika-XA – Advanced Analytics at Lower TCO

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Enterprise Requirements

• Reduced analytics footprint

• Superior performance for latency-sensitive analytics

• Out-of-the-box analytics engine, with flexibility to meet evolving needs

• Minimal management burdens

The Urika-XA Solution

• Single platform for wide range of analytic workloads

• Optimized for compute-heavy, memory-centric analytics

• Pre-integrated, tuned, and open platform

• Single point of support, scale compute-storage independently, compliant with enterprise standards

Page 8: Cray Urika-XA Advanced Analytics Platform

Urika Product Line

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Urika-XA • Extreme Analytics • Supports wide range of

analytic applications • Hadoop, Spark, and

future workloads • Batch and low-latency • Data mining, machine

learning, interactive data exploration

Urika-GD • Graph Discovery • Purpose built for discovery

analytics • Massively multithreaded

hardware accelerator to speed access to large, shared memory

• Graph representation, SPARQL query language

• Uncover hidden linkages and patterns

Page 9: Cray Urika-XA Advanced Analytics Platform

Why Cray?

• Emerging Analytics Needs: Require a new approach in order to deliver performance and lower TCO

• Advanced analytic techniques, data complexity, and time to value expectations are driving the need towards supercomputing-class architectures

• Established Expertise: Cray is THE supercomputer company, focused on developing data-intensive, low-latency technologies for over 40 years

• Pioneering use of fast interconnects, memory-centric architectures, system and workload management at scale

• Real-world use cases and deployments: We have a proven track record delivering high-performance, production-ready platforms for the most advanced analytics challenges

• Multiple mission-critical, production deployments in government, telecommunications, life sciences, financial services, and academia

• Use cases covering the spectrum of analytic needs in hard sciences and engineering

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Page 10: Cray Urika-XA Advanced Analytics Platform

Sample Use Cases

CRAY CONFIDENTIAL – DO NOT DISTRIBUTE

Financial Services

Risk Measurement

Life Sciences

Next-gen Sequencing

Government

Pattern of Life

Sports

Matchup Optimization

Telecom

Churn Analysis

Media

Data-driven Journalism