Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Isilon IQ
Scale-Out NAS Comes of Age
By Terri McClure
With John McKnight and Steve Duplessie
September, 2008
Copyright
ESG REPORT
Scale-Out NAS Comes of Age
Table of Contents
Table of Contents..................................................................................................................................................... i Scale-Out NAS Comes of Age ............................................................................................................................... 1 New Market Dynamics .......................................................................................................................................... 1 Addressing the Challenge: .................................................................................................................................... 2 Scale-Out File Storage ........................................................................................................................................... 2 Scale-Up versus Scale-Out ................................................................................................................................... 2 Scale-Up ............................................................................................................................................................... 2 Scale-Out .............................................................................................................................................................. 3 Scale-Out File Storage Attributes ......................................................................................................................... 4 Clustering .............................................................................................................................................................. 4 N-Way Clustering Approaches .............................................................................................................................. 4 Global Namespace-enabled ................................................................................................................................. 5 Power, Cooling and Space Efficiency (PCSE) ...................................................................................................... 5 Self-Managing and Self-Healing ........................................................................................................................... 5 Advanced Scale-Out Features ............................................................................................................................... 5 Transparent Data Mobility ..................................................................................................................................... 5 Tiered Storage Support ......................................................................................................................................... 6 Scale-Out NAS Comes of Age: .............................................................................................................................. 6 Isilon IQ .................................................................................................................................................................... 6 Isilon Advantage: SMP Architecture ..................................................................................................................... 7 Summary .................................................................................................................................................................. 8
All trademark names are property of their respective companies. Information contained in this publication has been obtained by sources The Enterprise Strategy Group (ESG) considers to be reliable but is not warranted by ESG. This publication may contain opinions of ESG, which are subject to change from time to time. This publication is copyrighted by The Enterprise Strategy Group, Inc. Any reproduction or redistribution of this publication, in whole or in part, whether in hard-copy format, electronically, or otherwise to persons not authorized to receive it, without the express consent of the Enterprise Strategy Group, Inc., is in violation of U.S. copyright law and will be subject to an action for civil damages and, if applicable, criminal prosecution. Should you have any questions, please contact ESG Client Relations at (508) 482-0188. This ESG White Paper was developed with the assistance and funding of Isilon Systems.
-iCopyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
80,000,000 70,000,000 60,000,000 50,000,000 40,000,000 30,000,000 20,000,000 10,000,000 0 Unstructured Database E-mail 2008 10,443,868 1,837,780 1,442,346 2009 15,808,970 2,991,043 2,557,446 2010 24,242,857 4,823,578 4,380,761 2011 39,364,875 8,110,447 7,745,201 2012 62,749,188 13,639,302 13,484,097
Source: ESG Research Report: Digital Archiving: End-User Survey & Market Forecast 2006-2010, January, 2006
-1Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
Not surprisingly, the massive growth of file data is driving growth in the Network Attached Storage (NAS) market. Vendors have been adapting technology to help users cope with managing file data growth; introducing denser NAS arrays, data reduction technologies, storage management, and storage optimization solutions. Like most technologies, the solutions were brought to market to solve an existing problemand most look backward rather than to the future. Today, the challenge with most enterprises is that file data growth is already out of control; this pattern of file data growth outpacing e-mail and database-driven growth has been going on for quite a while! Now, commercial enterprises are struggling with the new file characteristics of Internet Era data, further exacerbating the problem. It is no surprise that we often hear data center managers say that they love their first NAS applianceand curse their tenth, or worse yet, their hundredth! The growth of file-based data has left enterprise data centers bursting at the seams. The growth of file data, as well as the shifting nature of the files themselves to richer formats, is leading data center managers to consider taking a new approach to storing and managing file-based data and NAS vendors to introduce entirely new architectures. For managing growth and meeting the performance characteristics required by richer file data, scale-out is the buzzword of the day: the next step in NASs rich history of evolving to solve file storage and management challenges.
Scale-Up
Scale-up storage is just what it sounds like; it is designed to be monolithic, where lots of storage sits behind one or two file server heads and is designed to scale into the multi-TB range behind those file server heads. Once the limit on storage is hit, a new monolithic system is installed with a new file system to manage. There is no way to share the workload between the systems, and migrating directories or files between systems means remapping and remounting for each and every client with access. Those that have been through it know the pain of the process; it can be excruciating in a large enterprise environment with lots of clients and zero tolerance for downtime. Scale-up systems have no economical way to scale performance without some significant price penalty. Performance in todays monolithic systems is often scaled by adding a storage rack and more spindles to increase throughput and reduce latency (and, as a byproduct, reduce storage utilization). This is an expensive proposition for serving large sequential files. Adding processing power independently, as can be done with scaleout systems, not only saves floor and rack space. In addition to getting better performance, it would significantly reduce power consumption since processors typically use 95% less power than an additional disk shelf would consume.
-2Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
Scale-Out
Scale-out file storage utilizing standard NAS protocols (NFS, CIFS) meets the need for independent scale of storage capacity, processors, and bandwidth. Adding capacity and bandwidth, as well as file system expansion, is done online with minimal system performance impact. This granular scaling capability provides a price/performance advantage as it allows users to start small and scale where needed.
TABLE 1. SCALE-OUT VERSUS SCALE-UP NAS
Scale-out NAS meets a real market requirement for efficiently dealing with the large files typical of Internet Era file-based unstructured data. Recent ESG research indicates that scale-out NAS will be the fastest-growing segment of the file storage market (in both revenue and capacity) between 2007 and 2012, reaching 6.7 Exabytes in 2012 (see Figure 2).
FIGURE 2. SCALE-OUT NAS SHIPMENT FORECAST THROUGH 2012
-3Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
Clustering
A clustered file system runs concurrently on multiple physical storage nodes and is managed as a single entity. Essentially, a cluster removes the limitations of individual devices, thereby removing the boundaries of the boxes and enabling efficient management of multiple file servers. There are a number of approaches to clustering on the market. One approach is to employ clustering on a traditional scale-up architecture using a dual-node system. Commonly referred to as two-way clustering, dual node systems are primarily deployed for failover and to maintain high availability. Typically, these solutions enable one controller head to assume the identity of the failing controller head, and allow the failed controllers data volumes to continue to be accessed or written to by the new controller head. This inherently limits performance and scalability, as processing power is halved when one controller head fails. Management complexity and relative high cost to achieve the high availability are the main limiting factors with this approach. Unlike scale-ups two-way clustering implementation, scale-out systems employ n-way clustering that can start with as few as three nodes, but scales well beyond. The advantages of scale-out clustered systems are scale and ease of use. I/O loads are handled in parallel, leveraging distributed lock management and distributed metadata so any processing node is able to handle any request. Another advantage to clustered and independently scaled systems is the cost. Users can start out small and then grow into a massively parallel system. The performance ceiling is raised by adding more processors, the capacity by adding more storage for just-in-time scalability. And they can be easily managed because the entire cluster is handled as a single entity. IT managers simply cannot afford to manage hundreds of file systems individuallypeople dont scale.
Symmetric Clustered Architecture: Symmetric clustered architectures share many attributes of DFS clusters: symmetric clustered architectures grow resources seamlessly and enable the modular growth, or pay-as-you-grow, benefits of the storage system. When more memory, bandwidth, capacity, or drive
-4Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
actuators are needed, the cluster can be grown by simply adding additional nodes to the cluster. And symmetric clustered architectures provide extremely high levels of availability. But rather than leveraging a peer design, as with a DFS cluster, in symmetric cluster architectures as more nodes are added to the cluster, it still has one logical brain, regardless of the number of nodes in the solution. It maintains its coherency as one logical, dynamically expandable system. .
Global Namespace-enabled
This is a simple concept that is extremely difficult to achieve. In laymans terms, a global namespace is a virtual representation of a group of disparate physical file systems. It sits between clients and the assorted file servers in a given environment and adds a layer of abstraction that divorces what the client sees as mount points from the physical server mount points. It is a map that takes care of translating the virtual mount points to physical file servers and presents users with one consolidated view of the file server ecosystem. It is the secret sauce that enables a single point of management and advanced features, such as non-disruptive data migration and load balancing. It is important to differentiate native global namespace support from namespace aggregation. Namespace aggregation solutions essentially present a single pane of glass for administering storage management for multiple NAS systems. These solutions create gateways (either software-only or switch-based software) through which data from several different file systems is redirected to be accessed from a common point. Namespace aggregation solutions can typically control laying out a file (striping data) across disk volumes to a specific silo but not across the silos that make up the clusterwhile still allowing data movement between tiers of storage with limited or no client interruption. While this architecture approach can be attractive on the surface, the IT administrator is still managing, growing, and configuring islands of storage (heterogeneous silos of storage) but now with an additional virtualization layer. Ultimately, this solution approach can create higher complexity, higher management burden, and higher long term operational costs.
ESG REPORT
Scale-Out NAS Comes of Age
associated storage. This is another feature that some scale-out NAS vendors have implemented and are ahead of the curve onand is on the roadmaps of the rest.
For more information see ESG Brief: A Methodology for Driving Total IT Efficiency Using Four Simple Data Lifecycle Stages, June 2008 -6Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
Scale performance, bandwidth, and capacity independently: Isilon IQ provides granular scalability through its modular design. Performance is scaled by adding Isilon IQ Accelerator nodes, which add processing power, memory, bandwidth, and parallel read and write access to a single file system. Users can choose to scale single stream throughput and aggregate throughput or IO/s simply by adding more nodes. Isilon now ships 10 GbE support with the Accelerator-x product. Capacity can also be scaled by adding Isilon IQ-X storage nodes or Isilon EX storage expansion nodes. Self managing/transparent data mobility: Isilon IQ comes with a web-based management interface for single level management across the cluster. When nodes are added to the cluster, one click of the mouse (or front panel LCD) is required. The rest is automated. Isilons AutoBalance absorbs new storage into the cluster and grows the file system, rebalancing loads across cluster utilizing new nodes. And SmartConnect provides a single virtual host name for client mounts, then manages the distribution of client connections across the cluster based on defined policies. Power, Cooling, and Space Efficiency (PCSE): As stated previously, scale-out NAS systems are inherently power efficient because of their granularity of scale and right sizing scale. In other words, there is no need to add more spindles and consume energy on spinning rust to boost performance when a processor node can be added and use 95% less power. Isilon is also incorporating more power efficient components into the design, leveraging the power consumption efficiencies gained with the next-generation hardware advancements from Intel processors and power supplies. Isilon's X-Series achieves 20% greater power efficiency over Isilons previous architecture. Self-healing: Isilons FlexProtect data protection technology allows users to set data protection policies on the fly at an extremely granular level: cluster, directory, or file. Policies can be based on the desired level of data protection. Isilons N+4 protection also allows for up to four simultaneous failures without ever losing datano other storage system can with stand four failures like this in a single file system/volume. FlexProtect also delivers fast data rebuilds in the event of a drive or even full node failuredata can be rebuilt across any free space within the cluster, so space is not lost to spare recovery drives and recovery is extremely quick. Because Isilon OneFS can leverage all the nodes and spindles in a cluster to rebuild in the background, thus achieving massively parallel operationsa failed drive can typically be rebuilt as a background process in less than an hour. Tiered storage support: SyncIQ is a disk to disk replication product, but also supports simple, policybased file migration between storage tiers based on a number of characteristics, such as last access time, file name, or age. Entire directories or sub-directories can be included or excluded from migration jobs. This ensures only specific portions of the "source" file systemOneFSare migrated from online to nearline storage. Isilon also has all the features users have come to expect from traditional NAS systems, such as support for industry standards like NFS, CIFS, HTTP, FTP, NDMP, SNMP, LDAP, ADS and NIS; quota management; thin provisioning; and snapshots, and has an added layer of protection with its FlexProtect RAID support (N+1 through N+4).
-7Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
SMP is hard. Isilon was one of the first to market with an SMP-based clustered NAS solution. Isilon introduced quad core processors in the X-Series platform in January of 2008, but users were not able to realize the full impact until OneFS 5.0 was released as Isilon was only leveraging a single core. The newest release of OneFS unlocks the other three cores, and the SMP-based architecture automatically incorporates the cores into the workload sharing algorithms. But its not only X-Series customers that can realize the performance boost. Isilons clustered architecture is designed for both backward and forward compatibilityprevious versions of its processor and storage nodes can co-exist in a cluster with current versions. For users, that means aggregate cluster performance can be boosted by introducing new processor nodes into the cluster. The cluster absorbs the new capacity and automatically balances the load across the new cores. This is an important point and a clear advantagein a scale-up world, this kind of upgrade would require a forklift and a whole new system.
Summary
Isilon is in the sweet spot for new rich media and file-based data opportunities as the market is moving in its direction. Many traditional NAS vendors were late to recognize the shift and are just entering the scale-out market, while Isilon is already on its fifth generation product, giving it valuable experience. Isilon products are road tested and in use at leading companies like NBC Sportswhich stored video of the Beijing Olympic Games on Isilon systems for proxy and broadband contentproviding NBC producers with reliable access to critical content for rapid review, identification, and selection operations necessary to quickly produce and deliver groundbreaking coverage of the Beijing Olympics in the United States. If a companys success is measured in customer retention, NBC Sports use of Isilon speaks volumes; this is the third Olympics event where NBC has teamed with Isilon (Athens and Torino being the previous two). For a company few people know of, the client list reads like a whos who for rich digital content names: UCLA Laboratory of Neurological Imaging (LONI), The U.S. Geological Survey (USGS), Kodak EasyShare Gallery, NASA, NPR, Sony Music, ABC, Facebook, Second Life, MySpace, Paramount Digital Entertainment the list goes on and includes entertainment, oil and gas, Web 2.0, medical, and life sciences companies. Isilons focus has been to leverage its foot-hold in the HPC, Internet, media and entertainment markets into enterprise environments. The company is realizing success with this strategy. Isilon sits at the intersection where Web 2.0 meets business. Web 2.0 shares many HPC attributes: large files, scale-out being more important than scale-up, and a need for granular scale at the processor, bandwidth, file system, and storage capacity levelsonline and independently. As with other mission-critical systems, there is typically zero tolerance for downtime. While specialty file serving appliances have clear benefits in transactional IT environments, rich media is a whole new game where parallel file systems, clustering, and global namespace capabilities are of increased importance. Isilons big challenge is to draw the parallels and get commercial enterprises to realize the similarities. Dont take this to mean the incumbent players are ignoring the Internet-fueled market opportunity and challenges. Large incumbent vendors will not ignore this opportunity, but face the challenge of addressing these new requirements while maintaining their positions in other markets. This is the window of opportunity for Isilon and others to make a name for themselves.
-8Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.
ESG REPORT
Scale-Out NAS Comes of Age
-9Copyright 2008, The Enterprise Strategy Group, Inc. All Rights Reserved.