BigMemory & Hybris : Working together to improve the e-commerce customer experience



Similar documents
BigMemory: Providing competitive advantage through in-memory data management

terracotta technical whitepaper:

WITH BIGMEMORY WEBMETHODS. Introduction

BigMemory and Hadoop: Powering the Real-time Intelligent Enterprise

Ditch the Disk: Designing a High-Performance In-Memory Architecture

Big Data Management. What s Holding Back Real-time Big Data Analysis?

Using an In-Memory Data Grid for Near Real-Time Data Analysis

I N T E R S Y S T E M S W H I T E P A P E R INTERSYSTEMS CACHÉ AS AN ALTERNATIVE TO IN-MEMORY DATABASES. David Kaaret InterSystems Corporation

Microsoft SQL Server 2008 R2 Enterprise Edition and Microsoft SharePoint Server 2010

HyperQ DR Replication White Paper. The Easy Way to Protect Your Data

Microsoft Big Data Solutions. Anar Taghiyev P-TSP

IBM WebSphere Distributed Caching Products

Using In-Memory Computing to Simplify Big Data Analytics

Top 10 reasons your ecommerce site will fail during peak periods

Dell s SAP HANA Appliance

Tableau Server 7.0 scalability

How To Store Data On An Ocora Nosql Database On A Flash Memory Device On A Microsoft Flash Memory 2 (Iomemory)


Accelerating Enterprise Applications and Reducing TCO with SanDisk ZetaScale Software

JBoss Data Grid Performance Study Comparing Java HotSpot to Azul Zing

Contents Introduction... 5 Deployment Considerations... 9 Deployment Architectures... 11

Discover how customers are taking a radical leap forward with flash

Developing Scalable Java Applications with Cacheonix

Flash Performance for Oracle RAC with PCIe Shared Storage A Revolutionary Oracle RAC Architecture

Enterprise Edition Scalability. ecommerce Framework Built to Scale Reading Time: 10 minutes

An Oracle White Paper July Oracle Primavera Contract Management, Business Intelligence Publisher Edition-Sizing Guide

Accelerating the path to SAP BW powered by SAP HANA

CASE STUDY: Oracle TimesTen In-Memory Database and Shared Disk HA Implementation at Instance level. -ORACLE TIMESTEN 11gR1

The Flash-Transformed Financial Data Center. Jean S. Bozman Enterprise Solutions Manager, Enterprise Storage Solutions Corporation August 6, 2014

Ground up Introduction to In-Memory Data (Grids)

The Microsoft Large Mailbox Vision

Intelligent Business Operations and Big Data Software AG. All rights reserved.

VERITAS Storage Foundation 4.3 for Windows

SiteCelerate white paper

Protect SAP HANA Based on SUSE Linux Enterprise Server with SEP sesam

ScaleArc for SQL Server

Achieving Zero Downtime and Accelerating Performance for WordPress

Modern IT Operations Management. Why a New Approach is Required, and How Boundary Delivers

Databricks. A Primer

Mission-Critical Java. An Oracle White Paper Updated October 2008

Software AG Fast Big Data Solutions. Come la gestione realtime dei dati abilita nuovi scenari di business per le Banche

Benchmarking Couchbase Server for Interactive Applications. By Alexey Diomin and Kirill Grigorchuk

The Shortcut Guide to Balancing Storage Costs and Performance with Hybrid Storage

Digital Business Platform for SAP

How In-Memory Data Grids Can Analyze Fast-Changing Data in Real Time

Application Performance Management for Enterprise Applications

Converged, Real-time Analytics Enabling Faster Decision Making and New Business Opportunities

Improve Business Productivity and User Experience with a SanDisk Powered SQL Server 2014 In-Memory OLTP Database

Achieving Real-Time Business Solutions Using Graph Database Technology and High Performance Networks

MEETS BIG DATA BIG IRON. New opportunities from a goldmine of data

EMC XtremSF: Delivering Next Generation Storage Performance for SQL Server

WITH A FUSION POWERED SQL SERVER 2014 IN-MEMORY OLTP DATABASE

Why Alerts Suck and Monitoring Solutions need to become Smarter

Reducing the Cost and Complexity of Business Continuity and Disaster Recovery for

Transforming ecommerce Big Data into Big Fast Data

All-Flash Arrays Weren t Built for Dynamic Environments. Here s Why... This whitepaper is based on content originally posted at

CISCO WIDE AREA APPLICATION SERVICES (WAAS) OPTIMIZATIONS FOR EMC AVAMAR

WHITE PAPER The Storage Holy Grail: Decoupling Performance from Capacity

IS IN-MEMORY COMPUTING MAKING THE MOVE TO PRIME TIME?

Databricks. A Primer

The Flash Transformed Data Center & the Unlimited Future of Flash John Scaramuzzo Sr. Vice President & General Manager, Enterprise Storage Solutions

Agility for the Digital Enterprise Get There Faster

How To Speed Up A Flash Flash Storage System With The Hyperq Memory Router

The Flash- Transformed Server Platform Maximizing Your Migration from Windows Server 2003 with a SanDisk Flash- enabled Server Platform

Managing the Performance of Cloud-Based Applications

Amazon Cloud Storage Options

How To Improve Your Communication With An Informatica Ultra Messaging Streaming Edition

Simplified Management With Hitachi Command Suite. By Hitachi Data Systems

SOFTWARE-DEFINED STORAGE IN ACTION

Enabling actionable business intelligence in an Aruba Mobility Network environment

Powerful Duo: MapR Big Data Analytics with Cisco ACI Network Switches

GigaSpaces Real-Time Analytics for Big Data

Tips and Tricks for Using Oracle TimesTen In-Memory Database in the Application Tier

All-Flash Arrays: Not Just for the Top Tier Anymore

Microsoft SQL Server 2014 Fast Track

SSD Performance Tips: Avoid The Write Cliff

Accelerating Web-Based SQL Server Applications with SafePeak Plug and Play Dynamic Database Caching

Glassfish Architecture.

HyperQ Remote Office White Paper

EMC Backup and Recovery for Microsoft SQL Server 2008 Enabled by EMC Celerra Unified Storage

Web Application Deployment in the Cloud Using Amazon Web Services From Infancy to Maturity

Monitoring Best Practices for COMMERCE

TIBCO ActiveSpaces Use Cases How in-memory computing supercharges your infrastructure

The Cloud Hosting Revolution: Learn How to Cut Costs and Eliminate Downtime with GlowHost's Cloud Hosting Services

SAP HANA SAP s In-Memory Database. Dr. Martin Kittel, SAP HANA Development January 16, 2013

MySQL és Hadoop mint Big Data platform (SQL + NoSQL = MySQL Cluster?!)

Server Consolidation with SQL Server 2008

Breaking the Storage Array Lifecycle with Cloud Storage

<Insert Picture Here> Oracle In-Memory Database Cache Overview

Big Fast Data Hadoop acceleration with Flash. June 2013

Accelerating Hadoop MapReduce Using an In-Memory Data Grid

Jitterbit Technical Overview : Salesforce

FlashSoft Software from SanDisk : Accelerating Virtual Infrastructures

Diablo and VMware TM powering SQL Server TM in Virtual SAN TM. A Diablo Technologies Whitepaper. May 2015

Cloud computing means happier customers

SQL Server 2012 Parallel Data Warehouse. Solution Brief

WHITE PAPER. Flash in the SAN Panacea or Placebo?

never 20X spike ClustrixDB 2nd Choxi (Formally nomorerack.com) Customer Success Story Reliability and Availability with fast growth in the cloud

Upgrading to Microsoft SQL Server 2008 R2 from Microsoft SQL Server 2008, SQL Server 2005, and SQL Server 2000

Transcription:

& Hybris : Working together to improve the e-commerce customer experience TABLE OF CONTENTS 1 Introduction 1 Why in-memory? 2 Why is in-memory Important for an e-commerce environment? 2 Why? 3 How does Hybris work with? 5 Understanding performance gains 6 Managing scale with 7 Ready to learn more? This is the Age of the Customer. Customers want everything and they want it now. For an e-commerce setup, as data volumes grow, with more customers, more products and more orders, the performance gets impacted and the business struggles to retain customers or attract new ones. Thankfully, there is a simple solution. can maintain terabytes of data in-memory, which can include all customer information, the full product catalog and search results, among many other data sets, and guarantee low millisecond latency. Several sites are already benefiting from it and existing Hybris customers can also benefit. The integration is straightforward, and it takes only a few weeks to say goodbye to any performance issues, forever. Why in-memory? The combination of plummeting RAM prices and the rise of big data has become a gamechanger in the world of retail. While now retailers know their customers a lot better, they have to process more data and still support same or better performance. In-memory technology from overcomes this challenge by offering unlimited scale and guaranteed low latency. In-memory becomes more important in current times as data volumes have been measured to increase 10-fold every five years, quickly outpacing the capability of existing technologies and even Moore s law. Since both business and customer value are often associated with the need for massive amounts of data, enterprises are always looking for a solution. In-memory offers that solution by processing large data volumes in low milliseconds and analyzing data in real time to derive insights.

& Hybris: Working Together to Improve is the world s easiest, most powerful in-memory data management platform. makes your data available in real time to your customers, partners and employees, resulting in improved e-commerce performance, more satisfied customers and greater revenue. Why is in-memory important for an e-commerce environment? An e-commerce environment works with several different data types, such as customer information, the product catalog, orders and product promotions. Most of this data needs to be available instantly as any delays can impact the customer experience. It is even more important to ensure that there are no performance lags once a potential customer has entered the checkout flow to avoid losing out on the sale. Unfortunately, these objectives are not easy to achieve, primarily because: 1) All these different data types add up to several gigabytes of data (in most cases) and it is not possible to keep all of it in-memory without architecting a complex setup 2) As data grows, the in-memory capability needs to scale without impacting latency, as most systems do not scale in a linear fashion. This means that as the data volumes in-memory grow, the performance suffers 3) An in-memory system is an enabler; it enables faster transaction processing for an e-commerce environment. Hence, it should be easy to deploy and once deployed, easy to manage in the production environment. Not many systems are able to offer these two musthave features. With all data in-memory, the company can do real-time analysis to accelerate decision-making and satisfy customer requests much more quickly. This not only gives a potential competitive advantage and increased customer value, it also allows harnessing hidden value in the existing e-commerce data. As an example, an e-commerce business can analyze all data in real time by maintaining it in-memory and provide a personalized shopping experience to their customers. This can make the site stickier, drive additional traffic and increase overall conversion rate. Why? allows you to store data where it s used: in-memory. With, you can easily store and manage your in-memory data in a re-usable, standard way that simply plugs into your application, in this case Hybris. As a result, you can harness the value of massive amounts of data, with real-time processing, while keeping all of it in your server s RAM for maximum performance and scale. These performance improvements that result will continue to scale as your application s data set and user base grow. There s no simpler way to get predictable, and fast, access to large volumes of in-memory data. Without With 2GB 2GB Application SCALE UP Application Unused Memory 1TB 1TB COMMODITY SERVER COMMODITY SERVER By maximizing hardware utilization, can slash server hardware costs by up to 90 percent. 2

With, you get: & Hybris: Working Together to Improve Real-time access to terabytes of in-memory data High throughput with low, predictable latency Support for Java,.NET/C#, C++ applications 99.999 percent uptime Linear scale Data consistency guarantees across multiple servers Optimized data storage across RAM and SSD Microsoft SQL Server support for querying in-memory data Advanced monitoring, management and control Support for data replication across multiple data centers for disaster recovery employs a tiered approach to managing application data in-memory, automatically moving it between the different tiers as needed. The top two tiers the Java Virtual Machine (JVM ) heap memory and the in-process, off-heap store use the application server host s RAM. Since application server hardware typically ships with tens of gigabytes of RAM and can be inexpensively upgraded with hundreds of gigabytes or more, can efficiently store terabytes of data in RAM where your application can most readily use it. How does Hybris work with? Hybris is the world s leading e-commerce platform. It is built for large enterprises that process a lot of data in their e-commerce environment. Hence, it is not surprising to see that Hybris already uses in-memory/caching technology internally. But the caching is not set up to make 100 percent of data available for low millisecond access. This is how caching works in Hybris. can be used with the bottom two caching areas as shown in the boxes in green. Request Edge caching Akamai edge caching Edge caching enables storing data close to remote customer locations. This is done using a third-party service provider like Akamai that has a built in Content Distribution Network (CDN). Shop pages Page cache Page cache maintains page fragments in memory to make page loading time faster. For a large site, this can easily grow into hundreds of gigabytes of data and that s where can help. Persistence framework Hybris Persistence cache Persistence cache reviews the most frequently run queries and caches that data to make the customer get data faster. can fully replace this or integrate with it to make latency go down even further. Database Project-spec. Standard How Hybris Caching Can Benefit from can load the most commonly used data from the database into memory like product catalog, customers, orders, etc. This can be pre-loaded using s bulk loading feature to warm up the cache. 3

& Hybris: Working Together to Improve The figure below shows the topology of the s storage tiers. The top tier the Java heap memory in the application, JVM contains a maximum of two gigabytes of data, accessible in nanoseconds. The local tier the off-heap memory store in the application, JVM typically stores tens to hundreds of gigabytes of data accessible in microseconds. This directly integrates with Hybris as it sits on the same application server as Hybris. Finally, the distributed Tired Storage Latency Speed (TPS) Microseconds 2,000,000+ Tiered Memory Store 4 GB Process Memory App Server Hybris App Server Hybris Microseconds 1,000,000 32 GB - 12 TB Local Milliseconds 100,000 100s GB - 100s TB Distributed Server RAM or Flash/SSD Seconds 1 000s External Data Source (e.g. Database, Hadoop, Data Warehouse) Tiered Storage Model of array keeps hundreds of gigabytes to terabytes of data accessible to multiple application servers in milliseconds. Also shown is the speed of each tier for. For larger Hybris deployments with multiple application servers in a cluster, Hybris integrates with a cluster of servers to support maintaining hundreds of terabytes of data in-memory. This cluster has no limit on how big it can grow. Customers are already in production with several terabytes of data in-memory. The client interface maintains a TCP connection to the server array. The server array manages the movement of data between the different tiers as needed by the Hybris application. Server Array Stripe Stripe Stripe Stripe Stripe Stripe Stripe Database Ehcache API + Hybris Architecture 4

Understanding performance gains s tiered-store organization keeps data where an application like Hybris needs it for fast, predictable access precisely when it s needed. Because local memory is fast and increasingly cheap and abundant, keeps as much data locally as the available RAM permits. & Hybris: Working Together to Improve Two kinds of fast: high throughput and predictably low latency When measuring performance, it s tempting to look only at overall throughput the average number of transactions or operations per unit of time. But average throughput tells only part of the story. For instance, an e-commerce site may boast average throughput of thousands of requests per second, but outlier responses can take minutes. For applications where response time is critical page load time on an e-commerce site, for instance outlier response times are business killers. keeps applications safe from Java garbage collection by storing data in-memory, but not in the JVM heap. Applications can run with heaps small enough so the garbage collector never pauses the JVM, while keeping hundreds of gigabytes of data or more in memory. The result is a high throughput, low-latency system unburdened by long, unpredictable garbage collection pauses. Ultra-high availability: The five 9s is fast. It s also highly reliable, delivering 99.999 percent uptime. One reason is that s distributed architecture has no single points of failure. Each server in the array replicates data in real time to a mirror. Should a server go offline either for maintenance or due to unexpected hardware failure its mirror will replace it with zero down time. This is shown in the figure below. Stripe Server on commodity HW Transactionally updated mirror for high availability Active Server Mirror Server DRAM FRS DRAM FRS Stripe Server aka L2 is built for high availability. servers also protect themselves by throttling unexpected load spikes, giving the system the ability to adjust to changing usage patterns (rather than failing). And, built-in security measures prevent unauthorized access to data and services. Additionally, comes with a persistent store called the Fast Restartable Store (FRS) that enables fast recovery in case of a server crash. 5

& Hybris: Working Together to Improve s industry-standard API Applications access data through using the de facto Java standard, Ehcache API. It combines the simple get and put methods of a key-value store with powerful query, search and analysis capabilities, giving applications unprecedented visibility into data that might otherwise be locked away in slow, expensive disk-bound databases. stores data as plain Java objects. This simplifies programming and enables applications to efficiently use data without the overhead of the object-relational mapping transformation that comes with relational databases. Once data is in, it stays in the format most easily used by the application. also works well in heterogeneous technology environments. Because is deployed as an in-memory data service, customers commonly use MOM, HTTP, REST and SOAP protocols to access in a technology-agnostic way. Managing scale with is not only fast and reliable; it also helps effectively scale up and out over time to meet the rapidly evolving data requirements of any e-commerce environment. Linear scalability scales without bottlenecks along multiple dimensions. It scales up a single application server by maximizing hardware utilization and taking full advantage of the cheap, abundant memory on today s commodity hardware. scales out the application server tier through seamless data management across application servers. The server array scales out linearly, without impacting latency, providing ample headroom to support the growth of applications and data over time. Data consistency at scale When applications scale across multiple servers, it s crucial to manage data consistency between those servers. offers a range of consistency guarantees from strict XA-compliant transactions to eventual consistency all configurable on a per-dataset basis. Whether in a single JVM deployment or distributed across multiple application servers, manages data consistency over time according to your requirements. Monitoring, management and control comes with a full suite of monitoring, management and control tools and capabilities. s Automatic Resource Control (ARC) capability lets operators shape the allocation of memory on a per-dataset basis in each tier. When operators set a maximum memory allocation for a dataset in a tier, automatically maintains the size of that dataset within the allocation parameters, migrating data between tiers as necessary. ARC also offers a data-pinning option to guarantee that critical, frequently used data is always available in local memory. Through the management console, operators get run-time visibility into memory allocation at each tier and see application behavior and memory utilization, allowing them to make intelligent adjustments if necessary. also captures run-time statistics on your server topography, your server health, data access performance (by server and dataset) as well as remote JVM operating characteristics and thread dumps. All analytics are available through a number of formats, including log messages, a RESTful API, and our own management console for administration, monitoring, and management. You can also get statistics as JMX events for surfacing metrics into your own dashboards and monitoring software. 6

Configurable run-time events alert operators to changes in data store sizes, data access rates, and other performance indicators. Operators may then enable/disable data stores, adjust store sizes per tier by dataset, and adjust data freshness parameters. They can also control remote server life cycle to adjust data center topology and initiate remote backup procedures. & Hybris: Working Together to Improve s run-time visibility and alerting capabilities provide insight into what s happening in the data center. Operators can then use s remote server management and run-time control capabilities to take necessary action. Ready to learn more? is the easiest, most powerful way to take advantage of the in-memory revolution and make your Hybris environment overcome all performance and scalability challenges. With, you get fast, predictable access to all of your data up to hundreds of terabytes without garbage collection pauses. also gives all the reliability, availability and consistency that anyone expects from traditional disk-based data management systems. And most importantly, is easy to deploy and works without any issues, once in production. If you are ready to learn more, please contact your local Software AG representative or email us at tc-info@softwareag.com 7

& Hybris: Working Together to Improve Find out how to power up your Digital Enterprise at www.softwareag.com ABOUT SOFTWARE AG Software AG helps organizations achieve their business objectives faster. The company s big data, integration and business process technologies enable customers to drive operational efficiency, modernize their systems and optimize processes for smarter decisions and better service. Building on over 40 years of customer-centric innovation, the company is ranked as a leader in 14 market categories, fueled by core product families Adabas-Natural, Alfabet, Apama, ARIS, Terracotta and webmethods. Learn more at www.softwareag.com. 2015 Software AG. All rights reserved. Software AG and all Software AG products are either trademarks or registered trademarks of Software AG. Other product and company names mentioned herein may be the trademarks of their respective owners. SAG_Terracotta Hybris_8PG_WP_Sep15