Extreme Scalability and Flexibility for Access to 100% of ... · Oracle’s Sun ZFS Storage Appliance with multiple disk tiers • Oracle’s Pillar Axiom 600 storage system with

An Oracle Technical White Paper

February 2013

Extreme Scalability and Flexibility for Access to 100% of Your Data

Oracle Optimized Solution for Tiered Storage Infrastructure: Using StorageTek Storage Archive Manager to manage data on the Sun ZFS Storage Appliance, the Pillar Axiom 600 storage system, and StorageTek Modular tape systems.


Introduction ......................................................................................................... 1

Solution Objectives .............................................................................................. 2

Architecture Overview ......................................................................................... 3

Data Management ........................................................................................... 4

Tiered Storage ................................................................................................. 5

Server Infrastructure ...................................................................................... 12

Architecture Summary ................................................................................... 17

Configurations, Capacity, and Performance ...................................................... 17

Testing Tools ................................................................................................. 18

Test Results and Configurations .................................................................... 18

Best Practices for Configuring the Sun ZFS Storage 7420 Appliance and StorageTek SAM ........................................................................................... 23

Best Practices for Configuring the Pillar Axiom 600 Storage System and StorageTek SAM ........................................................................................... 24

StorageTek SAM Utilities and Performance ................................................... 26

Capacity Considerations ................................................................................ 27

Protecting New and Long-Term Data with StorageTek Data Integrity Validation ...................................................................................................................... 30

Conclusion ........................................................................................................ 33

References ........................................................................................................ 34


1

Introduction This paper focuses on how to deliver a scalable, flexible, and yet cost-effective tiered storage solution using the architecture and best practices defined in the Oracle Optimized Solution for Tiered Storage Infrastructure. This solution takes advantage of Oracle’s broad portfolio of storage, including all tiers from solid-state drives (SSDs) through tape, all managed by Oracle’s StorageTek Storage Archive Manager (StorageTek SAM). It also utilizes the compute power, security, and I/O features in Oracle’s SPARC T4 servers to provide a robust platform for managing unstructured content from small implementations to very large implementations with billions of files. The result is access to 100% of data in a cost-effective, scalable infrastructure.

The Oracle Optimized Solution for Tiered Storage Infrastructure is also designed to greatly simplify deployment and management and provide guidelines for component selection based on performance and capacity requirements. Additional information about the benefits of simplification and scalability for users as well as IT staff can be found in the companion business paper, “Making Data Archives Affordable Without Giving Up Performance or Capacity.”


2

Solution Objectives The Optimized Solution for Tiered Storage Infrastructure takes a storage infrastructure consisting of high-performance disk and SSDs, high-capacity disk, and tape and manages the data on that infrastructure with StorageTek SAM. It is designed to accomplish the following objectives:

• Increase storage efficiency

• Decrease overall storage cost over the lifetime of data

• Dynamically access data from any tier of storage enabling valuable information for collaboration and re-use

• Manage complex, time-consuming data management

• Gain control of explosive data growth

• Free up valuable IT staff cycles from working on low-value, manual data management tasks to higher-value, strategic and transformational projects

• Put long-term data automatically into the hands of the user

• Accelerate business decision making

• Enable successful product strategy decisions the first time based on historical data

• Re-use and share data with others with no time delay or time-consuming search and restore from a backup

• Survive IT transformations

• Ensure that valuable data can be accessed in the future regardless of changes in technology or IT staff

• Preclude vendor lock-in through open format, providing customers with more leverage in their supplier relationship


3

Architecture Overview

The Oracle Optimized Solution for Tiered Storage Infrastructure (shown in Figure 1) takes advantage of the perfect match of StorageTek SAM managing content on storage tiers that include

• Oracle’s Sun ZFS Storage Appliance with multiple disk tiers

• Oracle’s Pillar Axiom 600 storage system with multiple disk tiers and quality of service

• Oracle’s StorageTek Tape Library Systems with StorageTek LTO 6 or T10000C tape drives

This solution provides a very scalable, flexible, and cost-effective storage platform for many different use cases. This broad use demands an infrastructure that scales for both performance and capacity. Using Oracle's SPARC T4 server running Oracle Solaris delivers deployment flexibility and high performance. Oracle’s StorageTek modular library systems continue to provide long-term storage preservation enabling non-disruptive expansion.

Figure 1. The solution provides access to data from a multitier storage infrastructure that has been abstracted from the application through the use of StorageTek Storage Archive Manager, which manages data stored on primary disk, disk archive, and tape archive.

The architecture can be logically divided into the following three categories:

• Data Management—This is StorageTek Storage Archive Manager (StorageTek SAM), a powerful policy engine that enables customers to automatically move data to the appropriate storage tier based on access requirements.

• Tiered storage—This includes the storage devices consisting of primary disk, disk archive, and tape archive.

• Server infrastructure—This is the platform supporting StorageTek SAM software.

The following subsections provide an overview of these three major components of the architecture.


4

Data Management

StorageTek SAM is the critical component of the tiered storage because it ties together the different tiers while giving the application a simple, file-structured view of the data. It provides automatic and dynamic access to content from any storage device and presents a single file system to applications and users. Thus, StorageTek SAM hides the complexity of the tiered storage and provides transparent access regardless of where data is stored. Abstracting the storage from the application simplifies management, while providing all the benefits of scalability and flexibility across multiple storage tiers.

StorageTek Storage Archive Manager

StorageTek SAM is a storage software application that runs on Oracle Solaris. It requires Oracle Solaris Cluster for an active/passive high availability (HA) environment on the SPARC T4-1 or T4-2 servers. HA NFS enables applications to access data on the active node with the ability to fail over to the passive node if the active node fails.

Table 1 provides an overview of the components of StorageTek SAM and the specific version numbers used in the Oracle Optimized Solution for Tiered Storage Infrastructure.

TABLE 1. STORAGETEK STORAGE ARCHIVE MANAGER SOFTWARE COMPONENTS

SOFTWARE RELEASE

STORAGETEK SAM

SERVERS

Oracle Solaris Oracle Solaris 11.1

StorageTek Storage Archive Manager 5.3-U01

Oracle Solaris Cluster 4.1

StorageTek SAM accesses content from the primary StorageTek SAM disk cache based on preset policies and creates copies on archive disk and/or tape devices. StorageTek SAM can then dynamically access the content from any device. Up to four copies can be made locally and remotely, which provides a data-protection copy, eliminating the requirement for a backup. The remote copy is a remote disk archive, which is also a StorageTek SAM file system, allowing even more than the four copies to be created.


5

Each StorageTek SAM primary file system can scale to 4 PB in size; however, the capacity under StorageTek SAM management can reach hundreds of PBs through the use of tape media. The archiving file system policies automatically manage the lifecycle of the archive data through four features of StorageTek SAM:

• Archive—Transparently archives data from disk cache to archive disk and/or tape without operator intervention. The StorageTek SAM archiver uses policies based on file system characteristics, such as path name, wildcard, size, age, owner, group, or date to automatically manage the copies.

• Release—Automatically manages the disk cache and releases files from the disk cache that have been archived when the high-capacity threshold is reached on the primary storage or according to policy. The list of files eligible to be released is prioritized based on policies such as archive status, size, release status, and age. The inode remains on the primary disk and looks as if the data is located there as well.

• Stage—Automatically stages released files back to disk cache or directly to the requesting application when files are accessed. Staging options include prestaging and bypassing the disk cache. Removable media access is optimized for mounting and positioning.

• Recycle—Repacks archive media onto new media in order to reclaim space. The recycling process can be used to migrate from older to newer technology.

Throughout a file’s lifecycle, the StorageTek SAM metadata remains online and available to the content management application. All files appear to be directly located on the disk cache when they might be only on tape. The result is cost-effective management and use of tiered storage while providing dynamic and immediate direct access to 100 percent of the data without operator intervention or human knowledge about where the data resides. Thus, users have access to data that might be many years old or might not have been accessed in many years.

More information on StorageTek SAM can be found here:

http://www.oracle.com/us/products/servers-storage/storage/storage-software/storage-archive-manager/overview/index.html

Tiered Storage

Tiered storage is critical because content must be kept for long periods, yet some use cases require fast ingest as well as fast access for recently ingested data. Oracle has two disk storage products that meet the requirements of a tiered storage environment: both the Sun ZFS Storage Appliance and the Pillar Axiom 600 storage system are well suited to meet the requirements of the Oracle Optimized Solution Tiered Storage Infrastructure.

The Sun ZFS Storage 7420 appliance is used to store both the recently stored data and the most active data, no matter how old the data is. In addition, this storage also supports larger capacity storage for disk archive devices for StorageTek SAM. Flexible archive policies provided by StorageTek SAM keep copies of the data on the most appropriate storage tier. For smaller implementations, the Sun Storage 7320 ZFS appliance also meets high ingest and access requirements; therefore, it provides primary storage as well as disk archive for smaller configurations.




6

Oracle’s Pillar Axiom 600 storage system is used to hold the recently stored data and the most active data, and yet it can also provide disk archive in the same storage system. Flexible quality of service policies further aide in using this disk device in a multi-application environment.

Tape is used for archival of data that requires only occasional access, and it also provides data protection through multiple copies. The StorageTek T10000C includes Data Integrity Validation, based on ANSI standard CRCs. This provides additional data protection by validating that what was went to tape is what was written and that if inactivity or environmental factors have deteriorated the media, StorageTek SAM will be notified and a new archive copy will be created from an alternate copy.

Sun ZFS Storage Appliance

The Sun ZFS Storage 7420 appliance is an enterprise tier 1 storage system that radically simplifies and accelerates storage management. It features simple installation and administration, capacity scaling, and problem-solving capabilities with the integrated DTrace Analytics tool. It is ideal for enterprises requiring both high-performance storage as well as high capacity for disk archive, providing a dramatically easier and faster way to manage and scale the storage.

The Sun ZFS Storage 7420 appliance scales up to 1.15 PB of raw storage capacity and supports the Hybrid Storage Pool architecture, delivering the needed performance of ingesting data from sources such as high-capacity instrumentation. The combination of read- and write-optimized flash accelerators with high-performance hard disk drives (HDDs) delivers optimal performance for primary storage for StorageTek SAM, while the high-capacity HHDs meet the requirements of a disk archive. Figure 2 shows the combination of these disk architectures. This storage also provides both SAN and NAS interfaces to meet the requirements of StorageTek SAM as it manages multiple copies of the data on multiple tiers of storage.

In a tiered storage environment, for the highest tier, the Hybrid Storage Pool introduces an easy method of writing data to the correct high-speed storage without users or storage administrators making decisions. To deliver high performance, the Sun ZFS Storage Appliance file system—Oracle Solaris ZFS—seamlessly optimizes performance. It is designed to recognize I/O patterns automatically and place data on the best storage media using Hybrid Storage Pools.


7

Figure 2. Sun ZFS Storage 7420 appliance with two nodes and a tray with all HDDs and a tray with a mix of SSDs and HDDs delivering the flexibility required for a tiered storage environment.

For example, Oracle Solaris ZFS transparently executes writes to low-latency SSD media so that writes can be acknowledged quickly, allowing the application to continue processing. Oracle Solaris ZFS then automatically flushes the data to HDDs as a background task. This dynamic use of SSDs by the Sun ZFS Storage Appliance adds to the range of tiered storage that is managed by StorageTek SAM.

Configuration tasks are dramatically simplified in the Sun ZFS Storage Appliance through the browser user interface, which takes the guesswork out of system installation, configuration, and tuning. While testing the Oracle Optimized Solution Tiered Storage Infrastructure, the Sun ZFS Storage Appliance analytics were used extensively as different number of LUNs, different zpool configurations, different paths, and different StorageTek SAM configuration parameters were used.

More information on the Sun ZFS Storage Appliance can be found here:

http://www.oracle.com/us/products/servers-storage/storage/nas/overview/index.html

Pillar Axiom 600 Storage System

Pillar Axiom 600 is a modular enterprise-class storage system that provides multiple levels of performance and capacity. This makes the Pillar Axiom 600 storage system a good fit for content management, which involves very active content with possible high ingest rates as well as content that is older but still has high value and is accessed less frequently.

http://www.oracle.com/us/products/servers-storage/storage/nas/overview/index.html#http://www.oracle.com/us/products/servers-storage/storage/nas/overview/index.html


8

In this optimized solution, the recommended best practice is to use three classes of data stored on separate LUNs in the Pillar Axiom 600 storage system. StorageTek SAM metadata is stored in the highest-priority storage for fastest access, and the two additional storage categories are for StorageTek SAM primary storage and StorageTek SAM archive storage. These categories of data are further defined below and mapped in Figure 3.

• Premium level: StorageTek SAM metadata, database metadata, and database redo logs, as shown in red.

• High level: For the StorageTek SAM primary disk, as shown in green.

• Medium, low, or archive levels: For StorageTek SAM disk archive, as shown in blue.

Figure 3. The combination of I/O prioritization, multiple storage classes, I/O profiles customized to specific requirements, and ease of use results in the delivery of deterministic performance under any load condition.

It is this flexibility that makes the Pillar Axiom 600 storage system an excellent storage solution with StorageTek SAM for managing unstructured data. A single storage system with the three storage types and five priority levels of service delivers all tiers of disk storage in a single storage system with a single management tool.

Also represented in Figure 3 is the patented Pillar Axiom Quality of Service technology, which provides a big differentiator over traditional controller-based disk storage. Quality of Service (QoS) is delivered by prioritizing data access and ingest for different LUNs based on an assigned level of business priority. Advanced QoS software manages system resources (CPU, cache, and capacity) to automate storage provisioning based on business priority. Figure 3 shows what determines QoS for the components of the Oracle Optimized Solution for Tiered Storage Infrastructure that provide ingest, search, and access of content.


9

Pillar Axiom 600 Storage System Architecture

The Pillar Axiom 600 storage system is designed to scale performance along with capacity. Unlike most storage systems, which have a fixed number of storage controllers (usually a maximum of two), the Pillar Axiom 600 storage system can be scaled in multiple dimensions by independently adding more storage controllers or more trays of disks and/or SSDs as needed. The Pillar Axiom 600 storage system is built on three intelligent hardware assemblies, as described below and shown in Figure 4.

• Pillar Axiom Slammers (storage controllers):

• Two control units per Pillar Axiom Fibre Channel SAN Slammer

• Up to four active-active slammers

• Pillar Axiom Fibre Channel Brick, Pillar Axiom SATA Brick, and Pillar Axiom SSD Brick drive enclosures:

• One redundant RAID controller per brick

• Twelve FC drives, 13 serial ATA (SATA) drives, or 13 SSDs per brick

• Up to 64 bricks

• Up to 832 drives with a maximum of 1.6 PB

• Pillar Axiom Pilot (management platform):

• Easy-to-use user interface

• A single interface for managing physical and virtual configurations

Figure 4. Pillar Axiom 600 storage system components.

Slammers and bricks can be flexibly combined to meet unique application performance and storage capacity requirements. This flexibility is especially valuable to an application that takes advantage of tiered storage.

More information on the Pillar Axiom 600 storage system can be found here:

http://www.oracle.com/us/products/servers-storage/storage/san/pillar/pillar-axiom-600/overview/index.html




10

StorageTek Modular Library Systems

Oracle’s StorageTek tape libraries, tape virtualization, tape drives, tape media, and tape device software contribute to byte-stream preservation in a content management solution. Tape provides the highest level of availability at the lowest cost. Content can be stored for years in multiple locations without the use of power.

These systems also provide a copy (local and remote) while the content remains on spinning disk, delivering data protection and eliminating the requirement for a backup utility. The StorageTek modular library systems that are proposed in the small, medium, and large tiered storage solution scale from 30 to 100,000 slots, meeting all capacity, archive, and access requirements.

TABLE 2. LIBRARIES AT A GLANCE

STORAGETEK SL150 MODULAR

LIBRARY SYSTEM


LIBRARY SYSTEM


LIBRARY SYSTEM

NUMBER OF CARTRIDGE SLOTS 30 to300 200 to 5,925 1450 to 100,000

STORAGETEK T10000C CAPACITY n/a 1,000 TB to 29,626 TB 7,250 TB to 500,000 TB

STORAGETEK LTO 6 CAPACITY 75 TB to 750 TB 500 TB to 14,812 TB 3625 TB to 250,000 TB

MAXIMUM NUMBER OF TAPE

DRIVES 20 56 640

MAXIMUM NATIVE THROUGHPUT

(TB/HR) 11.5 48.4 552.9

SUPPORTED TAPE DRIVES HP LTO 6/5 StorageTek T10000C, StorageTek T10000B, StorageTek T10000A, StorageTek T9840D, and StorageTek T9840C

HP and IBM LTO 6/5/4/3

StorageTek T10000C, StorageTek T10000B, StorageTek T10000A, StorageTek T9840D, and StorageTek T9840C

HP and IBM LTO 6/5/4/3/2

NUMBER OF PHYSICAL PARTITIONS 2 8 8


11

Table 3 provides a description of the features of the two tape drives tested and recommended for the Oracle Optimized Solution for Tiered Storage Infrastructure.

TABLE 3. STORAGETEK TAPE DRIVES AT A GLANCE

STORAGETEK T10000C TAPE DRIVE STORAGETEK LTO 6 TAPE DRIVE

MEDIA CAPACITY 5 TB 2.5 TB

THROUGHPUT 252 Mb/sec 160 Mb/sec

NUMBER OF GENERATIONS OF MEDIA

SUPPORT

3 3

DATA INTEGRITY VALIDATION Yes1 Yes2

Selection of a library and tape drives for the Oracle Optimized Solution for Tiered Storage Infrastructure is made through the review of customer requirements, which include the following:

• Retention period of the content, which contributes to capacity requirements

• Number of copies on tape (a minimum of two is recommended), which contributes to capacity requirements

• Current content capacity to be archived

• Daily content capacity to be archived, which contributes to performance requirements

• Estimated yearly growth of content

• Other applications that will share the library

1 StorageTek T10000C uses CRC as defined in ANSI X3.139 and is calculated in the chip for no impact to performance.

2 LTO uses the Reed Solomon CRC calculation in software at an 88% decrease in performance. This is supported in IBM Tivoli.

REDUNDANT COMPONENTS Control path, fans, power Robotics, electronics, control path cartridge access ports (CAPS), fans, power

Robotics, electronics, control path CAPS, fans, power

REDUNDENT HOT-SWAPPABLE

COMPONENTS? yes yes yes

ANY CARTRIDGE ANY SLOT? yes yes yes


12

• Estimated activity of staging from tape to disk

• Customer requires Data Integrity Validation

More information on Oracle’s StorageTek modular library systems and tape drives can be found at

http://www.oracle.com/goto/tape

Management Applications

The following management applications are deployed separately from StorageTek SAM:

• Oracle's StorageTek Tape Analytics software proactively monitors StorageTek library environments to ensure the health and data availability of the global tape infrastructure. StorageTek Tape Analytics software is changing the way the world manages tape by moving from a reactive approach to a proactive, predictive approach. This software captures library, drive, and media health metrics in its dedicated server database and runs analytical calculations on these data elements to produce proactive recommendations for tape storage administrators. A proactive approach to managing the health of a tape environment improves the performance and reliability of existing tape investments.

• Oracle’s StorageTek Automated Cartridge System Library Software Manager is used to configure and control the tape drives.

Server Infrastructure

The server used in the Oracle Optimized Solution for Tiered Storage Infrastructure is based on the SPARC T4 series of servers. The SPARC T4 processor is the industry’s most highly integrated “system on a chip,” supplying the most high-performance threads of any multicore processor available and integrating all key system functions. Each SPARC T4 processor provides eight cores, with each core supporting up to eight threads (64 threads per processor). In addition, each core provides two integer execution pipelines, so a single SPARC core is capable of executing two threads at a time.

The SPARC T4 core architecture was redesigned, using features that are usually associated with superscalar designs that include out-of-order instruction execution, sophisticated branch prediction, prefetching of instructions and data, deeper pipelines, levels of caches, support for a 2-GB page size, and multiple instruction dispatching. This redesign resulted in an improvement in single-thread execution, networking, and throughput performance. These improvements lend themselves to an application, such as the Oracle Optimized Solution for Tiered Storage Infrastructure, that demands very high throughput.

http://www.oracle.com/goto/tape


13

The server models recommended for the solution include the SPARC T4-1 server for the small configuration and the SPARC T4-2 server for the medium and large configurations. The purpose of moving to the SPARC T4-2 for the medium configuration, even though StorageTek SAM is running in four cores, is the storage capacity requirements for a sufficient number of PCI slots to support disk and tape.

TABLE 4. SPARC T4 SERVER MODELS AND CHARACTERISTICS

SERVER MODEL SPARC T4-1 SPARC T4-2 SPARC T4-4

SIZE (RACK UNITS) 2U 3U 5U

CORES/THREADS 8/64 16/128 32/256

FREQUENCY 2.85 GHz 2.85 GHz 3.0 GHz

MAX. MEMORY 256 GB 512 GB 1 TB

PCIE GEN2 SLOTS 6 10 16

1 GBE/10 GBE PORTS 4/2 4/4 4/8

MAX. HDDS 8 6 8

SERVICE PROCESSOR Oracle Integrated Lights Out

Manager

(Oracle ILOM)

Oracle ILOM Oracle ILOM

Additional information on the SPARC T4 servers can be found at the following Website:

http://www.oracle.com/us/products/servers-storage/servers/sparc-enterprise/t-series/overview/index.html




14

A Multithreaded Hypervisor

The SPARC T4 processor offers a multithreaded hypervisor—a small firmware layer that provides a stable virtual machine architecture that is tightly integrated with the processor. Corresponding layers of virtualization technology are built on top of the hypervisor. The strength of Oracle’s approach is that all the layers of the architecture are fully multithreaded, from the processor up through applications that use the fully threaded Java application model. In addition to the processor and hypervisor, Oracle provides fully multithreaded networking and the fully multithreaded Oracle Solaris ZFS file system.

Oracle VM Server for SPARC (previously called Sun Logical Domains or LDOMs), Oracle Solaris Zones, and multithreaded applications are able to have allocated exactly the resources they need. In this solution, StorageTek SAM is supported in an LDOM, giving flexibility to the number of threads allocated. For the small and medium configurations, four cores are adequate for StorageTek SAM, while eight cores are recommended for the large configuration. This flexible allocation of resources helps control license costs for applications that are based on number of cores. Resources are then available to run additional applications in the same physical server.

Listed below are configuration rules when StorageTek SAM is implemented in an Oracle VM Server for SPARC environment.

• Oracle Solaris Zones (formerly call Oracle Solaris Containers). Only one non-global zone per file system is supported. There are several configuration restrictions regarding this feature and the QFS Administration Guide should be read prior to using this feature. For Oracle Solaris Cluster, Oracle Real Application Clusters (Oracle RAC) is supported in a zone cluster.

• Oracle VM Server for SPARC is supported with SAM-QFS, with the following restrictions:

• A minimum of four cores should be assigned to the domain.

• The suggested minimum of RAM is 24 GB.

• SAM-QFS MDS must boot from a physical device and, therefore, it must have at least one PCI root complex.

• Disk I/O virtualization is not supported for LUNs that are used in a QFS file system.

• Network virtualization is supported.

• Tape devices must be attached via non-virtualized PCI slots attached to the SAM-QFS MDS server.

• QFS clients may boot from a virtualized disk; however, they still need a PCI root complex to access file system devices via PCI controllers (FC, SAS, and so on).


15

• Oracle Solaris Cluster has the following requirements:

• Virtual storage devices used by Shared QFS must be backed by whole SCSI FC LUN disk arrays.

• Virtual storage devices must not be shared with any other guest domain on the same server.

• Virtualized partial LUN disk arrays are not supported.

Figure 5 shows how the server resources are allocated to the StorageTek SAM server for the storage software and for Oracle Solaris Cluster.

Figure 5. LDOM allocation of cores to StorageTek SAM virtual server.

The test results shown in Figure 6 with a workload of ingesting 100K records indicate that the CPU running the StorageTek SAM software scales up to 8 cores. Performance is similar for 16 cores and then drops when run in 32 cores.

Figure 6. Compare ingest of 100K records utilizing LDOMs allocating 4, 8, 16, and 32 cores to StorageTek SAM software.

Additional information for configuring LDOMs for StorageTek SAM is documented in the implementation guide.


16

Oracle Solaris Cluster

In today’s global 24x7 economy keeping enterprise applications up and running is more important—and can be more complex—than ever. Tiered storage is normally associated with dark archives that are rarely accessed and often high availability is not considered a requirement. However, this archived data provides an advantage in product development and corporate strategy, and the new normal is continuous access to 100 percent of all data, including old data that has been digitized, indexed, and archived. Government regulations, corporate financial goals, and evolving requirements to address new opportunities mean IT systems need to be constantly available. This can be a challenge with today’s complex solution stacks and unique business requirements. Recovery time objectives (RTOs) must be determined to decide whether a cluster environment is required to meet RTOs. The testing of the Oracle Optimized Solution for Tiered Storage Infrastructure included Oracle Solaris Cluster testing with StorageTek SAM and included the cluster agents HA-SAM and HA-NFS to ensure the highest availability.

The following are a few bullet points that apply to this solution and summarize the key capabilities and benefits of Oracle Solaris Cluster in a tiered storage environment. This release of StorageTek SAM runs in an active/passive environment. Since StorageTek SAM is running in four cores for the small and medium configurations and in eight cores for the large configuration, the remaining resources can be utilized for additional applications.

• High availability framework. Software that detects node failures quickly and activates resources on another node in the cluster. The framework includes a Cluster Membership Monitor, which is a distributed set of algorithms and agents that exchange messages over the cluster interconnect to enforce a consistent membership view, synchronize reconfiguration, handle cluster partitioning, and help maintain full connectivity among all cluster members.

• Virtualization support. Oracle Solaris Cluster provides comprehensive support for the following Oracle virtualization software: Oracle Solaris Containers, Oracle VM Server for SPARC, and Dynamic Domains (available on Oracle’s SPARC Enterprise M-Series servers). This enables flexible HA for server consolidation efforts. Applications can run unmodified in a virtualized environment.

• Flexible storage support. Oracle Solaris Cluster can be used with different storage technologies such as FC, SCSI, iSCSI, and NAS storage on Oracle or non-Oracle storage. There is broad file system and volume manager support, too.

• High availability Oracle Solaris ZFS. With the virtually unlimited scalability of ZFS, Oracle Solaris Cluster offers a file system solution with exceptional availability, data integrity, and flexibility for growth.

• Component monitoring. Extensive monitoring of application processes, disk path integrity, and network availability is done. For example, all disk paths can be monitored and configured to automatically reboot a node in the event of a multiple-path failure.


17

• Failover and scalable agents. There are software programs that support Oracle or third-party applications to take full advantage of Oracle Solaris Cluster features.

The implementation guide for the Oracle Optimized Solution for Tiered Storage Infrastructure includes specific configuration instructions for StorageTek SAM in an Oracle Solaris Cluster environment.

The following link provides additional detailed information about Oracle Solaris Cluster:

http://www.oracle.com/technetwork/server-storage/solaris-cluster/documentation/index.html

Architecture Summary

The previous sections described the different components that are included in the proven Oracle Optimized Solution for Tiered Storage Infrastructure. These are best-of-breed products that were engineered to work well together and have been tested with known results for performance, scalability, and flexibility.

The following sections include the results of testing, suggested configurations, and expected performance along with best practices for configuration and connectivity that will provide a guideline for selecting an architecture to match today’s requirements and scale as requirements grow.

Configurations, Capacity, and Performance

Two main points must be taken into consideration when selecting the hardware configuration for the Oracle Optimized Solution for Tiered Storage Infrastructure. One is capacity and the second is performance. Ingest performance is important for use cases that have instrumentation that generates TBs of data per hour. Capacity is important for use cases that might have a smaller daily ingest requirement but have a very long retention period (often forever) for millions of files. All use cases have specific requirements in both performance and capacity. The purpose of the tests and the test results is to provide guidance for selecting a configuration that most closely meets current requirements.

The following sections provide both capacity and performance comparisons for small, medium, and large configurations of the Oracle Optimized Solution for Tiered Storage Infrastructure. Another purpose of the test results is to show the scalability of the solution as business needs grow and data capacity increases.

http://www.oracle.com/technetwork/server-storage/solaris-cluster/documentation/index.html


18

An example of how to scale is that the Sun ZFS Storage 7320 appliance, which is used in the small configuration, can be upgraded to the Sun ZFS Storage 7420 appliance without moving data; you simply replace the storage nodes and configure the Sun ZFS Storage 7420 appliance to access the data in-place. In addition, StorageTek SAM and the Pillar Axiom 600 storage system have the ability to migrate data from old technology to new technology non-disruptively, and Oracle’s tape system strategy is to provide read access for three generations of tape media from the newest tape drives. This gives businesses many years of data access without migrating the content and provides the ability to take advantage of the latest tape drive technology and media for new archives. As tape media becomes more dense and appealing to a data center, StorageTek SAM provides the tools to migrate the content to this new technology.

Testing Tools

Two test applications were used during this testing: File System Test (also called FS Test or fstest) and DVT which can be found in the StorageTek SAM tools.

The FS Test called Multiprocess File System Performance Test Tool (mpfstest), writes to a StorageTek SAM file system. mpfstest aims to test a file system’s basic functionality and I/O performance. Unlike fstest, which is a single process and a single-file test tool, mpfstest is a multiprocess file system performance test tool. It is able to generate a multiprocess workload to measure and report a file system’s read and write performance.

The following areas are tested and measured by mpfstest:

• Multiprocess write performance test for files of fixed size

• Multiprocess write performance test for files of random sizes within a given range

• Multiprocess read performance test for files of any size

The following information is reported:

• Write rates

• Read rates

Test Results and Configurations

The reported test results are based on writing directly to the StorageTek SAM file system, reporting the highest rates achieved with the configuration available. The performance is optimal for the number of paths configured, reflecting that the storage side of the controllers allows the ingest rates to run at optimal speed. Access from a server using the NFS or CIFS protocols will have varying results due to the performance characteristics of these protocols. Additional application servers can be implemented to add additional throughput as needed.


19

Overall Sun ZFS Storage Appliance and StorageTek SAM Test Results

The following results apply to the small, medium, and large configurations when using the Sun ZFS Storage Appliance as the primary storage for StorageTek SAM.

A very detailed Sun ZFS Storage Appliance and FC implementation guide called “Understanding the Use of Fibre Channel in the Sun ZFS Storage Appliance” can be found here:

http://www.oracle.com/technetwork/server-storage/sun-unified-storage/documentation/o12-019-fclun-7000-rs-1559284.pdf

Testing on a medium and large configuration indicated that for this workload of writing 100K files, a zpool that is created using SSD and 15K drives and configured as RAIDZ provides similar performance as the same combination of SSD and 15K drives using RAID 10; therefore, a RAIDZ environment is ideal for also achieving the best available capacity. Figure 7 supports this.

Figure 7. Comparing RAIDZ and RAID 10 for the workload of ingesting 100K records results in slightly better performance and a nice side effect of better disk usage with RAIDZ.

Small Configuration

The small configuration contains the following:

• Two servers for a high availability configuration

• SPARC T4-1 server

• One 8-Gb FC HBA for each server

• 64-GB DIMM for each server

• StorageTek Storage Archive Manager software license for four cores

• Oracle Solaris Cluster license for four cores

• Two SAN switches




20

• Disk storage

• Sun ZFS Storage 7320 appliance cluster

• 48-GB DIMM per node

• One 8-Gb FC HBA per node

• Two disk trays

• Forty 3-TB drives

• Four SSD write flash accelerators

• Tape storage

• One StorageTek SL150 modular tape library (includes 30 slots and 1 tape drive)

• One 30-slot expansion shelf

• Three additional StorageTek LTO 6 tape drives

Capacity can be increased non-disruptively, as required, to a maximum of 432 TB of raw storage. Hybrid Storage Pools utilizing SSDs are the StorageTek SAM primary storage. Zpools without SSDs are used for the disk archive.

Medium Configuration

The medium configuration contains the following. You have a choice of either the Sun ZFS Storage Appliance or the Pillar Axiom 600 storage system.



• Three 8-Gb FC HBAs for each server




• One 10-GbE switch

• Two SAN switches


21

• Disk storage: Sun ZFS Storage 7420 appliance



• Two 8-Gb FC HBAs per node

• Two disk trays

Two SSD write flash accelerators for each tray Twenty 600-GB 15-K drives for each tray

• Two disk trays

Twenty-four 3-TB drives for each tray

• Disk storage: Pillar Axiom 600 storage system

• One slammer

• Four 8-Gb FC ports

• Four FC bricks with twelve 600-GB drives each

• Four SATA bricks with thirteen 2-TB drives each

• Tape storage

• One StorageTek SL3000 modular tape library

• Six StorageTek T10000C tape drives

• 900 activated slots

• Tape management

• Oracle's StorageTek Automated Cartridge System Library Software and server

• StorageTek Tape Analytics software and server


22

Large Configuration

The large configuration contains the following. You have a choice of either the Sun ZFS Storage Appliance or the Pillar Axiom 600 storage system.



• Two 10-GbE cards

• Six 8-Gb FC HBAs for each server




• Two 10-GbE switches

• Two SAN Switches

• Disk storage: Sun ZFS Storage 7420 appliance



• Four 8-Gb FC HBAs per node

• Four disk trays

Four SSD write flash accelerators for each tray Twenty 600-GB drives for each tray

• Four disk trays

Twenty-four 3-TB drives for each tray • Twenty 600-GB 15-K drives

• Twenty-four 3-TB drives

• Disk storage: Pillar Axiom 600 storage system

• Two slammers

• Four 8-Gb FC ports for each slammer

• Eight FC bricks with twelve 600-GB drives each

• Eight SATA bricks with thirteen 2-TB drives each


23

• Tape storage

• One StorageTek SL8500 modular library system

• Twelve StorageTek T10000C tape drives

• 1,750 activated slots

• Tape management

• StorageTek Automated Cartridge System Library Software and server

• StorageTek Tape Analytics software and server

Best Practices for Configuring the Sun ZFS Storage 7420 Appliance and StorageTek SAM Figure 8 shows how the storage pools can be configured for the primary storage for StorageTek SAM.

Figure 8. Sun ZFS Storage Appliance pools and LUN allocation for StorageTek SAM primary storage.

Configuration settings for the content pools and for the metadata pools are as follows:

• Four pools with six disks from each tray

• Four data LUNs per pool

• 128kB record size

• secondarycache=none

• Throughput for direct writes to the QFS file system


24

• One metadata LUN per pool

• Metadata LUNs: 16kB record size

• logbias=latency

• secondary cache=all

Block storage is required for StorageTek SAM primary storage; therefore, LUNs on the Sun ZFS Storage Appliance are created and mapped to the server. StorageTek SAM is then used as a volume manager and configures 1 to 4 LUNs into a single small file system for the metadata and configures up to 16 LUNs into a single file system for the content.

Best Practices for Configuring the Pillar Axiom 600 Storage System and StorageTek SAM

All testing and test results were for the purpose of configuring StorageTek SAM and the Pillar Axiom 600 storage system for the Oracle Optimized Solution for Tiered Storage Infrastructure. The results can be used as a guideline for similar customer requirements. The tests were not intended to be full performance testing for the purpose of general StorageTek SAM and Pillar Axiom 600 storage system use or for pushing servers and storage to their maximum performance.

Pillar Axiom 600 Storage System Configuration

Figure 9 describes the physical and logical configuration of the Pillar Axiom 600 storage system. The logical components are created using the Pillar Axiom 600 user interface. The StorageTek SAM file system uses the option of metadata separation, meaning the metadata and content are on separate volumes. The StorageTek SAM metadata resides on one logical device and the primary content resides on a different logical device. This configuration enables storing the metadata, which is small in size, on the highest-performing storage and storing the primary content, which is the actual data, on the next highest-performing storage.

Figure 9. Pillar Axiom 600 storage system physical and logical configuration for best performance in StorageTek SAM.


25

Figure 10 shows how the logical LUNs to be used for the StorageTek SAM data files are mapped to the server though the slammers when two slammers are available. All LUNs are then configured into the StorageTek SAM file system and presented to the application as a single file system for access via NFS or CIFS or as a POSIX-compliant file system. The same logic is used for the metadata. If two LUNs are created, whether on one or two Pillar Axiom SSD Bricks, StorageTek SAM will use both LUNs for the metadata. Pillar Axiom Quality of Service will ensure the appropriate priority for each record processed.

Figure 10. Mapping multiple LUNs from the Pillar Axiom 600 storage domains—utilizing all slammer paths—to the StorageTek SAM server. All LUNs are used to create the StorageTek SAM file system.

Configuring LUNs on the Pillar Axiom 600 Storage System

When configuring the Pillar Axiom 600 storage system for use by StorageTek SAM, it is recommended that four bricks be allocated per storage domain. As the system scales to meet new capacity and/or performance requirements, add bricks in a quantity of four at a time and create a storage domain utilizing four new bricks. The previous section describes how and where the LUNs are created and how the StorageTek SAM file system is created from those LUNs. This section describes the best practices for configuring LUNs for StorageTek SAM and the Pillar Axiom 600 storage system.

• Number of data LUNs per file system: Within a storage domain of four bricks, several LUNs are created and mapped to the server running StorageTek SAM, utilizing all of the slammer paths on the back end between the disk controller and the slammer HBAs. StorageTek SAM then puts the LUNs back together into a single StorageTek SAM file system.


26

• Number of metadata LUNs: Metadata performance had the greatest impact when using the StorageTek SAM utilities. Current customer experience suggests artificially limiting the number of files in a single StorageTek SAM file system in order to complete running samfsdump in an acceptable amount of time. Testing the StorageTek SAM and Pillar Axiom 600 infrastructure indicated that when utilizing SSD devices for StorageTek SAM metadata, the best configuration is two SSD LUNs that are mapped to two slammers. These SSD LUNs can be on the same SSD device. This configuration allows all paths in the slammers to be utilized for metadata. Testing with four SSD LUNs mapped to all slammer paths did not result in better performance, so more is not better in this case. Having two SSD LUNs was optimal.

Configuring StorageTek SAM

The StorageTek SAM disk allocation unit (DAU) and stripe width are important for write performance. The DAU setting is the minimum amount of contiguous space that is used when a file is written. Each file system has its own DAU. The stripe width specifies the number of DAUs to be written to a single LUN before switching to the next LUN.

The basic internal stripe width of the Pillar Axiom 600 storage system is 640K, so it is optimal for writes to attempt to write 640K to eliminate the RAID 5 read-modify-write penalty. The ability of the Pillar Axiom 600 storage system ability to coalesce writes will help achieve this goal with smaller DAUs.

StorageTek SAM Utilities and Performance

Metadata performance had the greatest impact when using the StorageTek SAM utilities. Current customer experience suggests artificially limiting the number of files in a single StorageTek SAM file system in order to complete running samfsdump in an acceptable amount of time. The charts in Figure 11 show the time required to run the utilities based on the number of files in the StorageTek SAM file system. These charts indicate that it is best to keep the number of files in a file system between 150 million and 200 million. As you approach 200 million files, the samfsrestore and samfsck times increase, although on the Sun ZFS Storage 7420 appliance, these utilities complete in a shorter time and you could push that limit higher. There are also other reasons, such as data usage, for restricting a file system.


27

Figure 11. Time to run the StorageTek SAM utilities samfsdump, samfsck, and samfsrestore using 50M, 100M and 200M files.

To achieve the performance in the utilities, the key is to run parallel threads. For 50M files, 10 threads were used, for 100M file, 20 threads were used, and for 200M file, 40 threads were used. The goal was to have each thread process about 5M files. At the completion of all jobs, samfsrestore can be run in parallel, using the files created by the dump command, or the individual dump files can be concatenated together and a single samfsrestore can be run. The implementation guide has sample scripts and instructions.

Capacity Considerations

When capacity is calculated based on ingest rates running 24x7, 365 days per year, it quickly grows out of proportion to the projected 50-percent growth per year. Table 5 represents capacity of content with a retention of seven years based on the proven continuous ingest rates for 24 hours per day, 7 days per week, ingesting 3.2GB/sec. Following the “rule of thumb,” we will store 10 percent of the total content on primary disk for frequent access, 30 percent will be stored on disk archive, and 200 percent will be kept on tape, representing data protection copies as well as archive. This sample has reached capacity much higher than the 50-percent growth projected.


28

TABLE 5. SEVEN-YEAR TABLE USING MAXIMUM INGEST RATE RUNNING 24X7 365 DAYS PER YEAR

INGEST GB/SEC INGEST

CAPACITY PER

DAY (TB)

INGEST

CAPACITY PER

YEAR (PB)

CAPACITY IN 7

YEARS (PB)

10% PRIMARY

DISK

30% ARCHIVE

DISK

200% TAPE

ARCHIVE

3.20 270.00 96.24 673.7 67.4 202.1 1,347.4

More realistically, the ingest rates noted in Table 5 will be experienced only at peak times, not 365 days/year; however, they must be processed with little or no impact to users during those peak times. The total ingest capacity will more likely be a much smaller percentage of the totals shown. To choose the solution that meets both performance and capacity requirements, select the small, medium, or large configuration based on peak ingest and access rates required, and select total capacity with actual expected growth and retention time. The scalability and flexibility of the solution will grow the infrastructure to meet new requirements.

As an example of capacity requirements, Table 6 starts with 50 TB of capacity for a small configuration, 500 TB of capacity for a medium configuration, and 1,000 TB of capacity for a large configuration with an expected growth of 50 percent per year through seven years. As stated previously, estimated capacity for each tier of storage is 10 percent for primary disk, 30 percent for disk archive, and 200 percent for tape archive to provide data protection.

TABLE 6. SEVEN-YEAR CAPACITY GROWTH, INCREASING 50% PER YEAR

YEAR 1 YEAR 2 YEAR 3 YEAR 4 YEAR 5 YEAR 6 YEAR

7

SMALL CONFIGURATION’S

CAPACITY GROWTH FOR 7

YEARS

TOTAL CONTENT CAPACITY

(TB)

50 75 113 169 253 380 570

PRIMARY DISK (10%) (TB) 5 8 11 17 25 38 57

ARCHIVE DISK (30%) (TB) 15 23 34 51 76 114 171

ARCHIVE TAPE (200%) (TB) 100 150 225 338 506 759 1,139

MEDIUM

CONFIGURATION’S


YEARS


(TB) 500 750 1,125 1,688 2,531 3,797 5,695

PRIMARY DISK (10%) (TB) 50 75 113 169 253 380 570

ARCHIVE DISK (30%) (T\PB) 150 225 338 506 759 1,139 1,709

ARCHIVE TAPE (200%) (TB) 1,000 1,500 2,250 3,375 5,063 7,594 11,391


29

LARGE CONFIGURATION’S


YEARS


(PB) 1.0 1.5 2.3 3.4 5.1 7.6 11.4

PRIMARY DISK (10%) (PB) 0.1 0.2 0.2 0.3 0.5 0.8 1.1

ARCHIVE DISK (30%) (PB) 0.3 0.5 0.7 1.0 1.5 2.3 3.4

ARCHIVE TAPE (200%)(PB) 2.0 3.0 4.5 6.8 10.1 15.2 22.8

The graphs in Figure 12, Figure 13, and Figure 14 make it clear that even with dramatic growth in overall capacity, the primary disk storage tier, which is the most expensive tier, remains relatively small, thus keeping costs low. The most cost-effective storage—tape—carries the largest capacity.

Figure 12. Small capacity, starting with 500 TB and growing 50 percent per year.

Figure 13. Medium capacity, starting with 100 TB and growing 50 percent per year.


30

Figure 14. Large capacity, starting with 1 PB and growing 50 percent per year.

All requirements must be taken into consideration when selecting a small, medium, or large configuration, with performance and capacity being the primary place to start. The Oracle Optimized Solutions team has proven that the components work together and the test results provide guidelines for size selection. With StorageTek SAM, the Sun ZFS Storage Appliance or the Pillar Axiom storage system, and StorageTek tape systems providing tools for migrating data non-disruptively, it is possible to begin with one configuration size and easily and confidently move to the next size.

Protecting New and Long-Term Data with StorageTek Data Integrity Validation

It is critical to ensure that stored data has been recorded accurately and just as important to be sure it remains unchanged through its retention time. StorageTek Data Integrity Validation, on Oracle’s StorageTek T10000C tape drive, takes this one step further by validating CRC checksums generated at the host. This integrity check for write, read, and validate has the highest level of importance when storing data that might be kept forever and yet has low access requirements.

Historically, data written to tape is verified after it has been written or validated at the inefficient full file level. StorageTek Data Integrity Validation starts this process at the server on a record level and continues the CRC check throughout the write process until the data is written to tape media. The StorageTek T10000C will again validate the data at the drive and at the server when it is read.

Performance on the server during the validation-only step is not affected because the validation does not require data to be staged back to the server for the CRC check process. The validation is executed on the drive in the background, and the server is notified to take action only if an error is detected. This process ensures all media, even if it contains dark archive files and is rarely if ever accessed, will be loaded into the drive and read on a schedule, such as yearly or every six months. A migration to new media after many years of being stored in a library slot will not be the first time a file is read.


31

DIV Process for Write, Read, and Validate

The step-by-step write, read, and validate processes are as follows:

• Write steps (from server to the StorageTek T10000C media)

• File is written by the application to a StorageTek SAM file system.

• Policy states that the file should be archived on tape and the archiver goes through the archive process.

• StorageTek SAM calculates the 32-bit CRC specified in ANSI X3.139 on each 2-MB record.

• This 4-byte CRC is added to each 2-MB record (configurable size) as it is sent to the StorageTek T10000C tape drive.

• The StorageTek T10000C tape drive receives and recalculates the CRC and compares it to the CRC on the record.

No-Match = StorageTek T10000C sends a request to the server to resend the record Match = StorageTek T10000C writes the record to tape

• Read steps (from StorageTek T10000C to server)

• The server sends a request for the file.

• Media is loaded into the StorageTek T10000C tape drive.

• The StorageTek T10000C drive reads the records and recalculates the CRCs and compares them to the CRC attached to each record.

No-Match = StorageTek T10000C sends a message to StorageTek SAM, which will stage the data from a second archive file and schedule the file for re-archiving

Match = StorageTek T10000C sends the record to the server • Server receives the record and recalculates the CRC and compares it with the CRC attached to

record.

No-Match = StorageTek T10000C sends message to StorageTek SAM; StorageTek SAM will stage the data from a duplicate archive image and will re-archive

Match = StorageTek T10000C sends the record to the application


32

• Validate steps

• StorageTek SAM sends a command to validate a range of StorageTek T10000C media.

• Media is loaded into the StorageTek T10000C tape drive as a background task.

• The StorageTek T10000C tape drive recalculates the CRC and compares it to 4 bytes on the record.

No-Match = StorageTek T10000C sends a message to StorageTek SAM, which will stage the data from another archive image and re-archive

Match = StorageTek T10000C sends a success status to StorageTek SAM and unloads the media

Performance Implication of StorageTek Data Integrity Validation

Both SPARC- and x86-based servers have the option to generate the required CRCs in the chip versus in software, requiring very little processor overhead. Test results from driving 10 StorageTek T10000C tape drives at optimal speed prove this. Figure 15, which compares using StorageTek Data Integrity Validation to not using StorageTek Data Integrity Validation, shows not only that StorageTek Data Integrity Validation has zero impact on performance but also the scalability of adding StorageTek T10000C tape drives.

The test environment for all the performance results graphed in Figure 15 used three StorageTek T10000C tape drives per 8-Gb FC HBA port. Data were read from two dual-slammer Pillar Axiom 600s and written to tape in parallel. StorageTek Data Integrity Validation mode was turned on. The first graph shows the zero performance impact of StorageTek Data Integrity Validation on StorageTek SAM archiving to 10 StorageTek T10000C tape drives.

Figure 15. Writing to 10 StorageTek T10000C tape drives in parallel with StorageTek Data Integrity Validation mode turned on shows no performance impact.

As tape drives are added, the amount of data archived by StorageTek SAM scales perfectly with the number of tape drives, as Figure 16 shows.


33

Figure 16. Adding two StorageTek T10000C tape drives at a time provides perfect scalability.

Looking at the data in another context in Figure 17, in addition to perfect scalability, as new drives are added, there is no impact on the performance of the drives already running as long as the disk system can read the data as fast as the tape drives can write the data. A drop in performance is more a function of the disk system than the tape system.

Figure 17. Additional StorageTek T10000C tape drives do not affect the performance of any tape drive.

With vast amounts of critical data being stored digitally, it is essential that the content of the data remain unchanged during transfer from server to tape media and back as well as when stored for long periods of time. For legal and preservation purposes, the fixity of this data must be verified. Oracle’s StorageTek Data Integrity Validation on the StorageTek T10000C tape drive allows StorageTek SAM to use CRC checksums to ensure the integrity of archived data is preserved.

Conclusion

Designed to address the challenges of rapid data growth and data management challenges associated with active archiving, Oracle Optimized Solution for Tiered Storage Solution automates data management processes to help organizations save time and money. The solution employs powerful, policy-based storage tiering and automated data movement to increase storage efficiency while reducing risk of loss of data or loss of access to data. The Oracle Optimized Solution for Tiered Storage Solution optimizes storage efficiency by ensuring that data is always kept on the storage tier that best matches the current access and retrieval requirements of that data. This automatic migration of data across storage tiers more accurately aligns the current business value of the data to the cost of its storage.


34

References For more information, visit the Web resources listed in Table 7.

TABLE 7. WEB RESOURCES FOR FURTHER INFORMATION

WEB RESOURCE DESCRIPTION WEB RESOURCE URL

StorageTek SAM configuration and documentation http://www.oracle.com/us/products/servers-storage/storage/storage-

software/storage-archive-manager/overview/index.html

SPARC T4 servers http://www.oracle.com/technetwork/server-storage/sun-sparc-

enterprise/overview/index.html

Oracle Solaris operating system http://www.oracle.com/technetwork/server-storage/solaris/overview/index.html.

Oracle Solaris Cluster http://www.oracle.com/technetwork/server-storage/solaris-

cluster/overview/index.html

StorageTek tape libraries and drives http://www.oracle.com/technetwork/server-storage/sun-tape-

storage/overview/index.html

Oracle Optimized Solution for Tiered Storage Infrastructure

http://www.oracle.com/us/solutions/optimized-solutions-171609.html

http://www.oracle.com/technetwork/server-storage/hardware-

solutions/index.html



http://www.oracle.com/technetwork/server-storage/sun-sparc-enterprise/overview/index.html

http://www.oracle.com/technetwork/server-storage/sun-sparc-enterprise/overview/index.html

http://www.oracle.com/technetwork/server-storage/solaris/overview/index.html

http://www.oracle.com/technetwork/server-storage/solaris-cluster/overview/index.html

http://www.oracle.com/technetwork/server-storage/solaris-cluster/overview/index.html

http://www.oracle.com/technetwork/server-storage/sun-tape-storage/overview/index.html

http://www.oracle.com/technetwork/server-storage/sun-tape-storage/overview/index.html

http://www.oracle.com/us/solutions/optimized-solutions-171609.html

http://www.oracle.com/technetwork/server-storage/hardware-solutions/index.html

http://www.oracle.com/technetwork/server-storage/hardware-solutions/index.html


Extreme Scalability and Flexibility for Access to 100% of Your Data February 2013, Version 1.0 Author: Donna Harland

Oracle Corporation World Headquarters 500 Oracle Parkway Redwood Shores, CA 94065 U.S.A.

Worldwide Inquiries: Phone: +1.650.506.7000 Fax: +1.650.506.7200

oracle.com

Copyright © 2013, Oracle and/or its affiliates. All rights reserved. This document is provided for information purposes only and the contents hereof are subject to change without notice. This document is not warranted to be error-free, nor subject to any other warranties or conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or fitness for a particular purpose. We specifically disclaim any liability with respect to this document and no contractual obligations are formed either directly or indirectly by this document. This document may not be reproduced or transmitted in any form or by any means, electronic or mechanical, for any purpose, without our prior written permission.

Oracle and Java are registered trademarks of Oracle and/or its affiliates. Other names may be trademarks of their respective owners.

Intel and Intel Xeon are trademarks or registered trademarks of Intel Corporation. All SPARC trademarks are used under license and are trademarks or registered trademarks of SPARC International, Inc. AMD, Opteron, the AMD logo, and the AMD Opteron logo are trademarks or registered trademarks of Advanced Micro Devices. UNIX is a registered trademark of The Open Group. 0113

Documents

Extreme Scalability and Flexibility for Access to 100% of ... · Oracle’s Sun ZFS Storage Appliance with multiple disk tiers • Oracle’s Pillar Axiom 600 storage system with