Exadata Backup Synergy with Oracle ZS3 Series and Comparison with EMC Data Domain



Similar documents
An Oracle White Paper March Integrated High-Performance Disk-to-Disk Backup with the Oracle ZFS Storage ZS3-BA

INCREASING EFFICIENCY WITH EASY AND COMPREHENSIVE STORAGE MANAGEMENT

An Oracle White Paper November Backup and Recovery with Oracle s Sun ZFS Storage Appliances and Oracle Recovery Manager

Optimizing Storage for Better TCO in Oracle Environments. Part 1: Management INFOSTOR. Executive Brief

Backing up the Big Data Stack

EMC Data Domain Boost for Oracle Recovery Manager (RMAN)

BACKUP AND RECOVERY FOR ORACLE ENGINEERED SYSTEMS WITH ORACLE'S SUN ZFS BACKUP APPLIANCE

EMC DATA DOMAIN OPERATING SYSTEM

<Insert Picture Here> Refreshing Your Data Protection Environment with Next-Generation Architectures

WHITE PAPER Improving Storage Efficiencies with Data Deduplication and Compression

An Oracle White Paper October Realizing the Superior Value and Performance of Oracle ZFS Storage Appliance

Oracle Maximum Availability Architecture with Exadata Database Machine. Morana Kobal Butković Principal Sales Consultant Oracle Hrvatska

EMC DATA DOMAIN OPERATING SYSTEM

EMC Data Domain Boost for Oracle Recovery Manager (RMAN)

Backup and Recovery Solutions for Exadata. Ľubomír Vaňo Principal Sales Consultant

How To Use An Org Storage Zs3-Ba

Backup and Recovery Solutions for Exadata. Cor Beumer Storage Sales Specialist Oracle Nederland

Protect Microsoft Exchange databases, achieve long-term data retention

Next-Generation Data Protection

EMC BACKUP MEETS BIG DATA

StorageTek Virtual Library Extension (VLE) a Single Common-Management Tieredstorage

Protect Data... in the Cloud

Maximum performance, minimal risk for data warehousing

SYMANTEC NETBACKUP APPLIANCE FAMILY OVERVIEW BROCHURE. When you can do it simply, you can do it all.

Analyzing Big Data with Splunk A Cost Effective Storage Architecture and Solution

2009 Oracle Corporation 1

Business-centric Storage for small and medium-sized enterprises. How ETERNUS DX powered by Intel Xeon processors improves data management

The Best Network Attached Storage Choice for Oracle Database and Software Environments

Preview of Oracle Database 12c In-Memory Option. Copyright 2013, Oracle and/or its affiliates. All rights reserved.

Turnkey Deduplication Solution for the Enterprise

IBM TSM DISASTER RECOVERY BEST PRACTICES WITH EMC DATA DOMAIN DEDUPLICATION STORAGE

Overview: X5 Generation Database Machines

Protecting Microsoft SQL Server with an Integrated Dell / CommVault Solution. Database Solutions Engineering

SUN ORACLE DATABASE MACHINE

An Oracle White Paper May Exadata Smart Flash Cache and the Oracle Exadata Database Machine

EMC XtremSF: Delivering Next Generation Performance for Oracle Database

June Blade.org 2009 ALL RIGHTS RESERVED

Using HP StoreOnce Backup systems for Oracle database backups

Business-centric Storage for small and medium-sized enterprises. How ETERNUS DX powered by Intel Xeon processors improves data management

Oracle Exadata: The World s Fastest Database Machine Exadata Database Machine Architecture

Why Oracle Database Runs Best on Oracle Servers and Storage. Optimize the Performance of the World s #1 Enterprise Database.

SUN ORACLE EXADATA STORAGE SERVER

Redefining Microsoft SQL Server Data Management. PAS Specification

Actifio Big Data Director. Virtual Data Pipeline for Unstructured Data

EMC NETWORKER AND DATADOMAIN

SUN STORAGE F5100 FLASH ARRAY

Get Success in Passing Your Certification Exam at first attempt!

EMC Backup and Recovery for Microsoft SQL Server

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

Oracle Database - Engineered for Innovation. Sedat Zencirci Teknoloji Satış Danışmanlığı Direktörü Türkiye ve Orta Asya

EMC XtremSF: Delivering Next Generation Storage Performance for SQL Server

Optimizing Backup & Recovery Performance with Distributed Deduplication

Long term retention and archiving the challenges and the solution

Backup and Recovery Redesign with Deduplication

WHY DO I NEED FALCONSTOR OPTIMIZED BACKUP & DEDUPLICATION?

ExaGrid - A Backup and Data Deduplication appliance

SAS and Oracle: Big Data and Cloud Partnering Innovation Targets the Third Platform

We look beyond IT. Cloud Offerings

Protecting enterprise servers with StoreOnce and CommVault Simpana

EMC DATA DOMAIN PRODUCT OvERvIEW

Exadata Database Machine

How To Protect Data On Network Attached Storage (Nas) From Disaster

Running Oracle s PeopleSoft Human Capital Management on Oracle SuperCluster T5-8 O R A C L E W H I T E P A P E R L A S T U P D A T E D J U N E

EMC VFCACHE ACCELERATES ORACLE

Inge Os Sales Consulting Manager Oracle Norway

Demystifying Deduplication for Backup with the Dell DR4000

IBM Storwize V7000 Unified and Storwize V7000 storage systems

ZFS Storage Solutions for Unstructured Data Challenges

Nexenta Performance Scaling for Speed and Cost

EMC Data de-duplication not ONLY for IBM i

Express5800 Scalable Enterprise Server Reference Architecture. For NEC PCIe SSD Appliance for Microsoft SQL Server

Symantec NetBackup 5220

Cost Effective Backup with Deduplication. Copyright 2009 EMC Corporation. All rights reserved.

MaxDeploy Ready. Hyper- Converged Virtualization Solution. With SanDisk Fusion iomemory products

Archive Data Retention & Compliance. Solutions Integrated Storage Appliances. Management Optimized Storage & Migration

Removing Performance Bottlenecks in Databases with Red Hat Enterprise Linux and Violin Memory Flash Storage Arrays. Red Hat Performance Engineering

EMC Backup and Recovery for Microsoft SQL Server

SMB Direct for SQL Server and Private Cloud

EMC DATA DOMAIN EXTENDED RETENTION SOFTWARE: MEETING NEEDS FOR LONG-TERM RETENTION OF BACKUP DATA ON EMC DATA DOMAIN SYSTEMS

VERITAS Business Solutions. for DB2

Turbo Charge Your Data Protection Strategy

An Oracle White Paper April Siebel CRM Customer Order Management on Engineered Systems High-Performing Siebel Customer Order Management

VMware Virtual SAN Backup Using VMware vsphere Data Protection Advanced SEPTEMBER 2014

How To Backup With Ec Avamar

SQL Server Storage Best Practice Discussion Dell EqualLogic

An Oracle White Paper September Oracle Exadata Database Machine - Backup & Recovery Sizing: Tape Backups

Unitrends Recovery-Series: Addressing Enterprise-Class Data Protection

How To Get A Storage And Data Protection Solution For Virtualization

ACCELERATING YOUR IT TRANSFORMATION WITH EMC NEXT-GENERATION UNIFIED STORAGE AND BACKUP

VMware vsphere Data Protection

Business white paper. environments. The top 5 challenges and solutions for backup and recovery

EMC Business Continuity for Microsoft SQL Server 2008

An Oracle White Paper July Expanding the Storage Capabilities of the Oracle Database Appliance

Transcription:

White Paper Exadata Backup Synergy with Oracle ZS3 Series and Comparison with EMC Data Domain Josh Krischer January 2014 2014 Josh Krischer & Associates GmbH. All rights reserved. Reproduction of this publication in any form without prior written permission is forbidden. The information contained herein has been obtained from sources believed to be reliable. Josh Krischer & Associates GmbH disclaims all warranties as to the accuracy, completeness or adequacy of such information. Josh Krischer & Associates GmbH shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof. The reader assumes sole responsibility for the selection of these materials to achieve its intended results. The opinions expressed herein are subject to change without notice. All product names used and mentioned herein are the trademarks of their respective owners.

Contents Executive Summary... 3 Oracle Exadata Database Machine... 4 Oracle Recovery Manager... 5 Oracle ZFS Storage ZS3 Series... 5 Oracle ZFS Storage Appliance ZS3 Series Product Description... 5 Functionality... 6 Reliability, Availability and Data Integrity... 7 Performance... 8 Economics... 8 Case Studies... 9 EMC Data Domain Overview... 11 Data Domain Architecture...11 De-Duplication with Hashing-based Algorithm...11 Scalability and Upgrade Path...12 Data Domain Add-on Chargeable Features...12 Conclusions and Recommendations... 13 Appendix 1: Oracle ZFS Storage ZS3 Series Specifications.... 15 Appendix 2: Oracle ZFS Storage ZS3 Series Software.... 16 Appendix 3: Comparison Summary Between Oracle ZS3 Series and EMC Data Domain.... 17 Josh Krischer & Associates GmbH. All rights reserved. P a g e 2

Executive Summary The exponential growth in the amount of data stored for analysis creates new challenges for system and database administrators for effectively managing the backup and restore of this data in the 24/7 economy. Backup data requires more and more storage capacity, combined with the need for very quick restores, forces organizations to look for a solution that is quick and easy to deploy and straightforward to manage. The current requirements from a backup/recovery process are reliability, performance, easy management, flexibility and costs. For Oracle Exadata Database Machine, the Oracle ZFS Storage ZS3 Series provides a high performance backup solution that dramatically reduces Oracle Database backup and restore times at significantly lower costs than competitive products. As one of Oracle s application engineered storage solutions, the tailored-in Oracle ZS3 Series exploits the coengineering with Oracle Database and Oracle Exadata to deliver a level of synergy unavailable to competitive backup systems. As a result, users of Oracle Exadata with Oracle ZS3 Series storage do not need to ensure the compatibility or interoperability of the server, operating system, firmware, storage and networking, dramatically reducing integration and deployment time and greatly simplifying systems operation. This results in significant reductions in CapEx and OpEx and solidifies the Oracle ZS3 Series position as a superior backup and restore solution for Oracle Exadata. the tailored-in Oracle ZS3 Series exploits the coengineering with Oracle Database and Oracle Exadata to deliver a level of synergy unavailable to competitive backup systems A comparison of the Oracle ZS3 Series with the EMC Data Domain deduplication storage systems in backing up Oracle Database shows that the Oracle ZS3 Series provides a much better solution. The Oracle Series: Delivers better backup and restore performance due to higher processing power and direct connection with high speed, low latency InfiniBand (8x faster than Data Domain) Provides fast enterprise quality disks with better MTBF than Data Domain s SATA drives Supports all levels of Oracle s Hybrid Columnar Compression (only available with Oracle storage) for highly effective Oracle Database data reduction Ensures much higher scalability (6x more capacity than Data Domain) without forklift upgrades Provides better availability with its dual-controller clustered configuration, data integrity with checksum end-to-end error detection, and correction of silent data corruption In addition to these technical advantages, the Oracle ZS3 Series delivers several economical advantages which reduce the CapEx and OpEx of the installations: No additional backup server hardware and software are required Oracle Hybrid Columnar Compression (only available with Oracle storage) compresses data up to 50X, reducing the amount of storage capacity required by 3x-5x or more Josh Krischer & Associates GmbH. All rights reserved. P a g e 3

Compressed data on the ZS3 Series can be immediately leveraged for secondary uses such as application development, test, and QA without rehydration Efficient management and troubleshooting through a user-friendly GUI and sophisticated storage analytics software reduces administration time Fast performance increases IT productivity and ensures that RTO and RTO SLAs are met Oracle Exadata Database Machine The Oracle Exadata Database Machine is purpose-built to run the Oracle Database. It leverages industry-standard hardware and unique software algorithms to deliver higher performance for Online Transaction Processing (OLTP), Data Warehousing (DW), and consolidation of mixed workloads than competing systems at a lower cost. Exadata Database Machine is a turn-key solution that includes all the hardware needed to run the Oracle Database, including the database servers, storage servers and InfiniBand networking all pre-configured, pre-tuned, and pretested by Oracle. The Exadata Storage Server (Exadata storage or Exadata cells) is used as the storage for the Oracle Database in the Database Machine. It runs the Exadata Storage Server Software that provides the unique and powerful Exadata technology including features such as Smart Scan, Smart Flash Cache, Smart Flash Logging, IO Resource Manager, Storage Indexes and Hybrid Columnar Compression. Exadata Storage Expansion Racks can be used to add capacity and bandwidth to the system. The Exadata Database Machine offloads data intensive SQL operations into the Oracle Exadata Storage Servers. By doing that data filtering and processing occurs immediately and in parallel across all storage servers as data is read from disk. This storage offload reduces database server CPU consumption and reduces the amount of data moved between storage and database servers. Some other interesting features include: Exadata Smart Flash (each Exadata Storage Server includes 4 PCI flash cards) accelerates Oracle Database processing by speeding I/O operations. The Flash provides intelligent caching of database objects avoiding physical I/O operations. The Exadata Storage Server Software also provides the Exadata Smart Flash Logging feature to speed database log I/O. Hybrid Columnar Compression (HCC) - Exadata storage provides an advanced compression technology, that typically provides10x, and higher, levels of data compression and significantly improves the effective data transfer. The HCC technique utilizes a combination of both row and columnar methods for storing data and achieving the compression benefits of columnar storage, while avoiding the performance shortfalls of a pure columnar format. A logical construct called the compression unit stores a set of Hybrid Columnar-compressed rows as well. Queries run directly on Hybrid Columnar Compressed data and do not require the data to be decompressed. HCC is available for all Oracle engineered systems and Oracle ZS3 Series and Pillar Axiom storage when attached to Oracle Database. Josh Krischer & Associates GmbH. All rights reserved. P a g e 4

Oracle Recovery Manager The Oracle Recovery Manager (RMAN) utility is the Oracle technique for backing up and recovering the Oracle Database. It saves storage space and data transfer times by using file multiplexing and compression features. Oracle RMAN uses the incremental backup technique by backing up only the RMAN database blocks that have changed since the last backup. Oracle RMAN merges these changed blocks into the original image backup to create a new image of the Oracle data files which enables a full restore without the need to merge incremental backups into a full backup as part of the restore operation. This technique in addition to reducing the storage requirements also reduces significantly the backup and in particular the restore times. To ensure data integrity RMAN uses block-level corruption detection during backup and restore processes. Oracle RMAN offers data encryption capabilities as well as three levels of compression which optimize CPU utilization and the compression ratio to reduce storage capacity and network bandwidth requirements. Oracle RMAN can backup data to disk or tape. Oracle RMAN is a component of the Oracle Database. Therefore, there is no need for additional backup servers or extra software licenses to buy, and no third-party technology to purchase and manage. Oracle ZFS Storage ZS3 Series Oracle s ZS3 Series complements the extreme performance of Oracle Engineered Systems, including Oracle Exadata, with throughput of up to 26TB/hr for full backups and 17TB/hr for full restores -- a 30 percent increase for backups and 80 percent increase for restores over the previous generation. As result, the ZS3 Series significantly reduces backup and restore times for Oracle Exadata, ensuring that backup windows and Recovery Time Objectives (RTOs) are met for high demand SLAs. The tailored-in Oracle ZS3 Series is co-engineered with Oracle Database and Oracle Exadata exploiting the synergy among the three products to deliver a solution that is superior to competitive backup systems. The Oracle ZS3 Series is accessed via the NFS protocol. A faster protocol, Oracle Direct NFS (dnfs) significantly accelerates backup and restore performance by performing concurrent direct I/O, which bypasses any operating system level caches and eliminates any operating system write-ordering locks. In addition, dnfs Client performs asynchronous I/O, which allows processing to continue while the I/O request is submitted and processed. The Oracle ZS3 Series also supports NDMP enabling it to be equally effective as a backup target for Oracle Database installations running on Oracle SPARC T5, M5, and M6 systems as well as non-oracle servers, as well as providing backup to tape storage. Oracle ZFS Storage Appliance ZS3 Series Product Description The Oracle ZS3 Series is available in two models: the ZS3-4 and the ZS3-2. Both are available in single or dual-clustered controller configurations, and include a rich set of data services, the Hybrid Storage Pool intelligent cache architecture, multi-threaded SMP operating system, as well Josh Krischer & Associates GmbH. All rights reserved. P a g e 5

as a DRAM-centric system design and Oracle dnfs to power its superior backup and restore performance. The file system used is the advanced Oracle Solaris ZFS with 128-bit addressability. The ZS3-4 provides 2TB of DRAM, 12.7TB of read Flash cache, and 10.5TB of write Flash, scales to 3.5PB using 7.2K or 10K SAS-2 HDDs and is powered by 4 x 10 core 2.4GHz Intel Xeon processors per controller satisfying the current capacity and future backup requirements of most customers and ensuring investment protection (see Appendix 1 for complete system specification).the ZS3-2 provides 512GB of DRAM, 12.7TB of read Flash cache, and 10.5TB of write Flash, scales to 768TB using 7.2K or 10K SAS-2 HDDs and is powered by 4 x 8 core 2.1GHz Intel Xeon processors per controller. In both systems, data movement is optimized by sophisticated algorithms in the Hybrid Storage Pool architecture to ensure that 70-90% of data is read from DRAM, the fastest access method some 1,000 times faster than Flash. The Oracle ZS3 Series offers both disk-to-disk (D2D) and disk-to-disk-to-tape (D2D2T) backup connectivity options. It is established as a target for Oracle Recovery Manager (RMAN) backups, exploiting the advantages of Oracle Database and Oracle RMAN features to speed up data backup and restore processes. Host connectivity: Both ZS3 Series storage systems supports InfiniBand,1Gb and 10Gb Ethernet and 16Gb Fibre Channel connectivity options. InfiniBand connectivity, available in the ZS3 Series storage and used for high-speed direct connection to Oracle Exadata, is a unique feature among all backup appliances. The 10GbE and 16Gb FC options provide rapid access to Oracle StorageTek tape storage solutions for customers who want to add an extra layer of data protection or need long-term data archiving to meet regulatory requirements. Functionality The full table of software functionality is shown in Appendix 2.However, it is important to mention some unique functions. Oracle Hybrid Columnar Compression (HCC) is one of the best examples of the synergy among Oracle Database, Oracle Exadata and the Oracle ZS3 Series. HCC enables reduction in storage requirements for customers with existing NAS-based Oracle Database with in-database archives for OLTP, data warehousing, or mixed workloads. This unique functionality, not supported in its fullest form by any other storage platform, is a standard feature at no additional costs for users. With HCC, compression ratios up to 50x can be achieved, leading to storage capacity savings of 3x-5x over competitive systems, while query performance improves by 2x-5x. In addition to HCC, the Oracle ZS3 Series supports four levels of data compression 1 and in-line, block-level deduplication. 1 The 4 levels of compression differ by compression ratio and the required server overhead. The user can select the optimal level suiting his requirements. Josh Krischer & Associates GmbH. All rights reserved. P a g e 6

Reliability, Availability and Data Integrity The Oracle ZS3 Series uses several techniques to ensure availability, data integrity and to prevent data loss or silent data corruption. The dual-clustered controllers provide redundancy during maintenance operations or in the event of a complete controller outage, backup and restore operations can continue. The ZS3 Series supports end-to-end Checksumming which reads and compares data to ensure that it s correct. Several predictive and self-healing capabilities ensure system availability by automatically diagnosing, fencing, and recovering from faults. The system also provides detailed diagnostic messages that link to Oracle s knowledgebase, guiding administrators through corrective tasks when human intervention is required. In addition, the enterprise quality disk drives have better Mean Time Between Failures (MTBF) and the industry-leading triple-parity RAID further reduces the risk of data loss in particular for the large capacity NearLine disks. Disaster Recovery schemes can be deployed by using remote replication to another ZS3 Series storage system to protect against total system or site loss on the primary. Additional data protection can be provided by the D2D2T option by backing up to a tape library system, such as Oracle s StorageTek SL8500 modular library system, shown in Figure 1. The ZS3 Series maintains a complete backup on disk for fast restore times, and additional copies can be archived to tape. To reduce eventual performance degradation, the tape backups are initiated from Oracle RMAN and utilize the most recent backup file stored on the ZS3 Series to create a copy for the tape library. Oracle Exadata Oracle Oracle s StorageTek Database machine ZS3 Series Tape Library Josh Krischer & Associates GmbH. All rights reserved. P a g e 7

Performance The Oracle ZS3 Series storage systems feature an innovative architecture that maximizes I/O throughtput. This architecture includes an intelligent cache design based on Hybrid Storage Pools (HSPs) that provides very large DRAM and flash cache capacity, and enterprise-class disks; a multi-threaded SMP operating system that takes full advantage of the large number of CPU cores across two controllers in the ZS3-4 system. This architecture was the foundation for the ZS3-4 system s record setting performance on the SPC-2 benchmark with a world record (at the time) of 17,244.22 SPC-2 MBPS and the second best overall price-performance with a result of $22.53 SPC-2 price-performance. 2 The SPC-2 benchmark provides a source of comparative storage performance for streaming data workloads and is a good predictor of a system s backup and restore performance. Indeed, these robust capabilities coupled with the synergy with Oracle Exadata and Oracle Database enable the ZS3-4 system to achieve throughput of up to 26TB/hr for full backups and 17TB/hr for full restores a 30 percent increase for backups and 80 percent increase for restores over the previous generation. This backup rate is fast enough to completely backup an Oracle Exadata Database Machine X4-2 half rack configuration in less than 6 hours or a full rack configuration in less than 12 hours. Economics Oracle RMAN and Hybrid Columnar Compression are included with the Oracle Database and the Oracle ZS3 Series storage systems connect directly to Oracle Exadata's internally managed InfiniBand network, thus eliminating the need for a backup server hardware, software and the associated backup applications. This Oracle-on-Oracle solution reduces integration costs as well as the complexity and risk that come with managing multi-vendor systems. The Oracle ZS3 Series storage systems connect directly to Oracle Exadata's internally managed InfiniBand network, thus eliminating the need for a backup server hardware, software and the associated backup applications Simplified management reduces personnel costs through the use of an intuitive GUI that shrinks administration time by more than 30 percent by taking the guesswork out of system configuration, provisioning and tuning. In addition, Oracle s DTrace Analytics provides deep visibility for administrators to monitor crucial system parameters that can affect backup and restore of Oracle Exadata environments, speeding the resolution of performance bottlenecks and other issues. DTrace monitors, in real-time, the Oracle ZS3 Series processor utilization, cache usage, data transfers and other system-related data. Administrators can drill down to areas of concern to get more precise information which helps in problem resolution. These across the board savings significantly reduce the CaExp and the OpEx of the backup/restore solution based on Oracle Exadata and ZS3 storage systems in comparison to those of other vendors. 2 Results as of Sept. 10, 2013. Full results at http://www.storageperformance.org/results/benchmark_results_spc2#b00067. Josh Krischer & Associates GmbH. All rights reserved. P a g e 8

Case Studies The ZS3 is a new product therefore we were not able to publish case studies which include it. However, taking in consideration that the ZS3 is doubling the performance than its predecessors hints that these users could get even better benefits with the new product. Location: Texas, United State Industry: Professional Services Founded in 1998, Novation is a leading healthcare supply chain expertise, analytics, and contracting company for the more than 65,000 members of VHA Inc. and UHC, two national healthcare alliances, the Children's Hospital Association, an alliance of the nation s leading pediatric facilities, and Provista, LLC. Novation provides alliance members with sourcing services, as well as information and data services. The organization develops and manages competitive contracts with more than 600 suppliers. Many hospitals rely on the company s supply chain data management service to monitor monthly spend, analyze purchases, and help ensure an adequate stock of medical supplies. The company has seen its transaction volume growth rates double approximately once every six months (400% annually), requiring the company to increase database performance speeds, improve backup capabilities to limit system downtime, enhance analytics capabilities, and streamline its back-office infrastructure to provide information to its hospital and healthcare alliance customers more rapidly, while also enhancing internal business analytics. Solution: The company selected Oracle Exadata Database Machine to meet its growing database needs. Oracle Exadata helped make data processing times 10 to 15x faster than the legacy solution, and reduced transaction processing times from hours to minutes. The Oracle solution also enables the company to provide analytics to hospitals more quickly, supporting increased visibility into savings opportunities either in pricing or in contracting. The improved analytics speed also enables internal executives to react more quickly to market changes. The Oracle solution also allows the company to store more data on less hardware with the Oracle Hybrid Columnar Compression functionality, which provides significant savings while also increasing performance and enabling faster access to information. It also enabled the company to consolidate several applications on the Oracle Exadata Database Machine, simplifying database administration, infrastructure, and management. The company also selected Oracle s Sun ZFS Storage 7420 appliance as its backup storage device to enable very rapid data backup up to 6 terabytes per hour and help reduce system downtime. The InfiniBand connectivity between Oracle Exadata and the storage appliance enables Novation to rapidly move data between the devices and access the data as quickly as needed. Josh Krischer & Associates GmbH. All rights reserved. P a g e 9

Why Oracle "We selected Oracle because we needed a system that met our performance expectations, was more reliable, and reduced infrastructure redundancy to provide hospitals with the information they need as quickly as possible. The Oracle Exadata platform performed much faster than the competition. Further, Oracle s Sun ZFS Storage 7420 appliance enables us to securely backup our data while limiting downtime key requirements to meet our customers performance expectations, said Alex Latham, senior director, operational applications development, Novation The full case study can be found: http://www.oracle.com/us/corporate/customers/customersearch/novation-1-sun-sl-1969886.html FAIR Health, an independent not-for-profit corporation dedicated to bringing transparency to healthcare costs and out-of-network reimbursement, has selected the Oracle Exadata Database Machine with a cluster of Oracle s Sun ZFS Storage 7420 appliances for backup to support its database of more than 15 billion claims. An additional Sun ZFS Storage Appliance supports remote disaster recovery and development. FAIR Health s data are used by multiple stakeholders in health insurance, including insurance carriers, third party administrators, healthcare providers, employers, researchers and consumers to process claims, inform pricing models, offer cost transparency and to understand geographic variation in cost and utilization. FAIR Health sought a solution to streamline performance and eliminate the storage bottlenecks in its existing environment. With Oracle Exadata, FAIR Health tested its existing code and achieved significant performance improvements, including a 5x reduction in claims processing time(1) and a 10x faster average performance improvement in applying complex statistical methodologies(2). Using Oracle Exadata s Hybrid Columnar Compression technology combined with the Sun ZFS Storage Appliance, FAIR Health achieved an average of 14x reduction in physical storage space for its primary Oracle Database storage, database backups, and disaster recovery site. The Sun ZFS Storage Appliance s direct InfiniBand connection to the Oracle Exadata and its high-throughput design enables FAIR Health to reduce their backup windows and increase resources available for production workloads. FAIR Health also uses the Sun ZFS Storage Appliance s snapshot and replication capabilities to replicate compressed data to a third remote Sun ZFS Storage Appliance at its disaster recovery location where clones of the Hybrid Columnar Josh Krischer & Associates GmbH. All rights reserved. P a g e 10

Compressed data are used to develop proof-of-concepts for new application functionality. With the increased performance and capacity delivered by the Oracle Exadata and Sun ZFS Storage Appliance solution, FAIR Health will be able to reallocate resources to focus on product development and enhancement, including its expanded custom analytics capability. The full case study can be found: http://www.oracle.com/us/corporate/press/1897852 EMC Data Domain Overview EMC Data Domain is a family of general purpose storage systems whose claim to fame is in-line deduplication for backup and archiving. Data Domain supports all major backup applications, including Oracle RMAN. EMC discloses very few technical details about the Data Domain models. Internet search results provide mainly marketing information. There is no public information on which processors are used, cache size, Flash, RAID levels, etc. Data Domain Architecture All the Data Domain models are based on a single controller. This single controller represents a Single Point of Failure (SPOF). A component failure on the controller may cause a total system outage with potentially dire consequences, such as the inability to back up or restore data, and even data loss. The family includes 7 models from the entry level DD160 to the high end DD990. The models differ in performance and capacity. The specifications are shown in table 1. Table 1: Data Domain family (source EMC) The three low models and the DD990 use SATA disks and the three others use SATA and SAS. As stated above the Data Domain models are general purpose appliances that are not specially designed for nor have any unique integration points with Oracle Database or Oracle Exadata. DataDomain Management Center is a dashboard based virtual appliance which manages and monitor up to 75 Data Domain subsystems through a single interface. De-Duplication with Hashing-based Algorithm Hashing is CPU intensive and the hash tables must be kept in memory to maximize performance. A major problem with the hash-based algorithm is the very large index that it requires. If the repository grows to the extent that the hashing tables cannot be contained in memory, Josh Krischer & Associates GmbH. All rights reserved. P a g e 11

performance will drop dramatically. This can be observed in particular with low end models of the Data Domain family with less processing power (Intel Xeon dual-core CPUs) and less memory for the hash table. Scalability and Upgrade Path Data Domain upgrades are not smooth and require forklift upgrades, that is, physical replacement. To the best of our knowledge there is no technical upgrade option between the DD670 and the DD860, DD890 to DD990. Physical upgrade means data migrations, operation interruption and may interfere with amortization time. For government agencies, forklift upgrades may require issuing a new RFP. Data Domain Add-on Chargeable Features Data Domain Boost In-line de-duplication may suffer from poor performance under heavy load. To compensate, EMC introduced the Data Domain Boost (DD Boost) software, an agent which runs on and offloads some of the de-duplication process to backup or database production servers. These servers compress and send only unique data segments across the network to the Data Domain storage system speeding up back up and reducing networking bandwidth requirements, according to EMC. While the Data Domain Boost software compensates for the relative low processing power of the Data Domain single-processor controllers, why should users have to pay for additional licenses and experience slower application performance due to the CPU load caused by running deduplication on their servers to correct Data Domain s lack of performance due to a design flaw? Data Domain Extended Retention Although the largest usable capacity of the DD990 is 570TB, it can be extended by another chargeable feature called Data Domain Extended Retention. This feature creates two tiers of storage on the Data Domain storage system; tier 2 is most suitable for long-term data retention. Again, users are forced to pay to make up for another of Data Domain s design flaws: lack of capacity. Data Domain Performance As seen in Table 1, EMC claims that the DD990 top model can backup 15TB/hr without the chargeable DD Boost. This performance is much slower than the Oracle ZS3-4 system s 26TB/hr backup throughput. Backup is important but restore is much more important EMC hasn t published restore figures for the new Data Domain models. The Oracle ZS3-4 current delivers 17TB/hr of restore throughput. As for engineered systems environments, EMC has not published results since 2010. Those results were 2.7TB/hr for backup and 2.1TB/hr for restore with the Data Domain 880 3. Clearly not in the same class as the Oracle ZS3-4 system, or even the previous generation of the Oracle ZFS Storage Appliance. 3 http://www.emc.com/collateral/hardware/technical-documentation/h6835-backup-recovery-oracle-clariion-datadomain-ra.pdf Josh Krischer & Associates GmbH. All rights reserved. P a g e 12

Data Domain Deduplication in Oracle Database Environments EMC claims that Data Domain deduplication can reduce backup and archive storage capacity requirements by an average of 10-30 times. This figure may be achievable when backing up files such as documents, images, e-mails or Microsoft Share Point where duplicate files are prevalent. In contrast, relational databases, such as the Oracle Database, usually store data only once therefore the de-duplication factor is much lower. Further, Oracle RMAN uses a proprietary format which makes the backup stream largely opaque to third-party backup applications. This opaqueness combined with RMAN s own compression or HCC, leaves little duplicated data left for Data Domain to act on. In addition, deduplication is completely ineffective when data is encrypted by RMAN. In fact, EMC published a white paper (before HCC was available) titled EMC Backup and Recovery for Oracle 11g OLTP which shows a deduplication factor of only 6.3:1 in backing up Oracle Database to Data Domain. For maximum deduplication in backing up an Oracle Database, EMC recommends turning off HCC and performing full backups. In addition, data should not be encrypted, and archived log files should not be included. The bottom line is that compression via HCC is much more effective than deduplication in reducing Oracle Database data and it s only available on Oracle storage. Conclusions and Recommendations One of the current IT trends is the popularity of converged systems. Users demand simplicity, scalability, performance, manageability, and ease of use. Through its acquisition of Sun Microsystems, Oracle inherited Sun s server experience and Sun s StorageTek division, which can look back at 43 years of experience in storage technologies. The co-engineering among these groups has resulted in the synergy from which the Oracle Exadata, Oracle Database, Oracle RMAN and Oracle ZS3 Series emerged as an enterprise-grade backup/restore solution to protect the enterprises mission-critical data. This combination ensures high performance, lower capital and operational costs and can be deployed faster than the integration of a third party backup platform from another vendor. Customers further benefit from single-vendor support for all software and hardware components and avoid finger pointing in problem determination and resolution. A comparison of the Oracle ZS3 Series with the EMC Data Domain deduplication storage systems in backing up Oracle Database shows that the Oracle ZS3 Series provides a superior solution (a comparison summary is provided in Appendix 3). The Oracle ZS3 Series: Delivers better backup and restore performance due to higher processing power and direct connection with high speed, low latency InfiniBand (8x faster than Data Domain), Provides fast enterprise quality disks with better MTBF than Data Domain s SATA drives Supports Oracle Hybrid Columnar Compression (only available with Oracle storage) for highly effective Oracle Database data reduction Josh Krischer & Associates GmbH. All rights reserved. P a g e 13

Ensures much higher scalability (6x more capacity than Data Domain) without forklift upgrades Provides better availability with its dual-controller clustered configuration, data integrity with checksum end-to-end error detection technique, and correction of silent data corruption In addition to these technical advantages, the Oracle ZS3 Series delivers several economical advantages: No additional backup server hardware and software are required Oracle Hybrid Columnar Compression (only available with Oracle storage) compresses data up to 50x, reducing the amount of storage capacity required by 3x-5x Compressed data on the ZS3 Series can be immediately leveraged for secondary uses such as application development, test, and QA without rehydration Efficient management and troubleshooting through a user-friendly GUI and sophisticated storage analytics software reduces administration time The fast performance increases IT productivity and ensures that RTO and RTO SLAs are met Josh Krischer is an expert IT advisor with over 44 years of experience in highend computing, storage, disaster recovery, and data center consolidation. Currently working as an independent analyst at Josh Krischer & Associates GmbH, he was formerly a Research Vice President at Gartner, covering mainframes, enterprise servers and storage from 1998 until 2007. During his career at Gartner he was responsible for high-end storage-subsystems and disaster recovery techniques. He spoke on these topics and others at a multitude of worldwide IT events, including Gartner conferences and symposia, industry and educational conferences, as well as major vendor events. Find more on: www.joshkrischer.com Josh Krischer & Associates GmbH. All rights reserved. P a g e 14

Appendix 1: Oracle ZFS Storage ZS3 Series Specifications. Josh Krischer & Associates GmbH. All rights reserved. P a g e 15

Appendix 2: Oracle ZFS Storage ZS3 Series Software. Josh Krischer & Associates GmbH. All rights reserved. P a g e 16

Appendix 3: Comparison Summary Between Oracle ZS3 Series and EMC Data Domain. Oracle ZS3 Series Co-engineered for deep integration with Oracle Database, Oracle Exadata Database Machine and other Oracle Engineered Systems. Industry-leading performance with Oracle Engineered Systems, with backup and restore throughput rates of 26TB/hr and 17TB/hr respectively. ZS3-4 system s record setting performance on SPC-2 benchmark with a world record (at the time) of 17,244.22 SPC-2 MBPS and the second best overall price-performance with a result of $22.53 SPC-2 price-performance. 4 Native support for Hybrid Columnar Compression (HCC) for 10x-50x compression ratios and up to 8x faster query performance. Short backup and restore windows with highthroughput architecture, high-speed InfiniBand connectivity and optimized Direct NFS. Two controllers and a clustered architecture mean even a planned software upgrade won t take your backup and restore system offline. Backups can be used for development, test, or QA, with immediate and full access to HCC data from Oracle Recovery Manager (RMAN) images without the need for decompression. HCC 10x-50x compression and higher throughput performance reduces backup and secondary processing footprints. Superior performance and efficiency mean fewer systems are required lowering capital and operational costs. EMC Data Domain Not certified for backup with Oracle Engineered Systems. EMC has not published Oracle Engineered Systems backup results since 2010 those results were 2.7TB/hr backup and 2.1TB/hr restore. Not available or published HCC is not fully supported on EMC storage systems. Long backup and restore times for Oracle Engineered Systems. A single controller means all backups and restores will be unavailable in the event of a hardware failure or controller software upgrade. No ability to access or deduplicate HCC data in RMAN image backups. Inferior performance and compression leads to storage sprawl as more equipment is needed to meet capacity demand. Storage sprawl and numerous integration points mean more complexity and add up to higher CapEx and OpEx. 4 Results as of Sept. 10, 2013. Full results at http://www.storageperformance.org/results/benchmark_results_spc2#b00067. Josh Krischer & Associates GmbH. All rights reserved. P a g e 17