Holy Grail of Data Storage Management, The

Description

• Copyright 2000
• Dimensions: 7" x 9-1/4"
• Pages: 352
• Edition: 1st

• Book
• ISBN-10: 0-13-013055-9
• ISBN-13: 978-0-13-013055-6

1305E-9

"This is a great book at the right time . . . I found the book to be exactly what I was looking for and very well written." --Dr. David Spuler, Director of Advanced Research, BMC Software and Author, Enterprise Application Management with PATROL (1999)

What Every Enterprise Needs to Know to Solve Its Data Deluge!

Depending on the analyst one follows, corporate IT departments will spend between 75 and 90 cents of every dollar over the next five years on data storage products. The reason is simple: Companies are generating data at a phenomenal rate and increasing their requirements for data storage by 100 percent or more per year.

In The Holy Grail of Data Storage Management, Jon William Toigo documents current trends in storage technology and shows IT executives exactly how to plan a comprehensive strategy for maximizing the availability, performance, and cost-effectiveness of enterprise storage. Discover how to:

• Map your storage strategy to long-term business goals and application data movement requirements
• Apply architectural, scalability, and investment protection criteria to every storage purchase
• Customize storage to key enterprise applications, including data warehousing, ERP, OLTP, and e-commerce
• Master the new skills needed to manage next-generation storage

This vendor-neutral guide offers new insight into every next-generation storage technology: network attached storage (NAS), RAID array configurations, storage appliances, near on-line storage, Storage Area Networks (SANs), optical systems, and much more. If you're responsible for enterprise storage, planning, architecture, and/or distributed systems, you'll find this book absolutely indispensable.

The Holy Grail of Data Storage Management is complemented by a new website, http://www.stormgt.org, which provides useful, up-to-the-minute information on the fast-changing world of storage and storage management technology. See the Introduction for more details.



Sample Content

Downloadable Sample Chapter

Click here for a sample chapter for this book: 0130130559.pdf

Table of Contents

I. THEORY OF STORAGE MANAGEMENT.

1. Data Storage Trends beyond the Millennium: The Road to Chaos.

A Paradigm Shift. Storage Management Defined. Endnotes.

2. Why Storage Management?

Increased Risk of Data Loss. Increased Personnel Cost. Increased Storage Acquisition Costs. Increased Downtime due to Storage-Related Issues. Increased Downtime due to Absence of an Availability-Optimized Architecture. Decreased Performance in Applications. Diminished End User Confidence and Satisfaction. Impediments to Storage Management. Endnotes.

3. Storage in the Modern Distributed Enterprise.

Data Distribution Models Evolve. Distributed-Anarchical Data Distribution. Regionalized-Managed Data Distribution. Centralized-Managed Data Distribution. A Continuum of Options. Endnotes.

II. STORAGE TECHNOLOGIES.

4. The Perpetuation of a Disk-Based Storage Strategy.

Evolution of Hard Disk Technology. Enabling Technology for Disk Drive Advancement. Read-Write Heads. Other Drive Improvements. New Initiatives in Drive Technology. PRML Channel Technology and Enhanced Management Features. Interfaces: Connecting Storage for Efficient Use. Advanced Technology Attachment/Intelligent Disk Electronics (ATA/IDE) Interface. Small Computer System Interface (SCSI). Ultra SCSI. Serial Storage Architecture (SSA). Fibre Channel. Thinking Outside the Box. Endnotes.

5. Raid.

Disk Storage Sub-systems: Cutting the Tether. RAID Levels. Beyond RAID 5. Implementing RAID. Stand-Alone Arrays. RAID Objectives Scorecard. Endnotes.

6. The Advent of SANs.

Fibre Channel: A SCSI Replacement? Networking Concepts for Storage Managers. SANS: Problems Solved, Issues Remain. Endnotes.

7. Network-Attached Storage (NAS) and Storage Appliances.

Distributed Storage Using NAS. Defining Terms. NAS and the Thin-Server Revolution. The NAS Operating System. Differences in Kernel Implementation. Anatomy of a NAS Operating System. Homogeneous NAS File Systems. Enhancing File Systems and MultiProtocol File Access. A Kludge at Best. Integrating NAS with the Enterprise Storage Infrastructure. Endnotes.

8. Secondary Storage: New Roles for Tape.

The Market for Tape. Tape in the Large Systems Market. Tape Products Crowd the Mid-Range. DLT: King of the Mountain. Knocking DLT of the Top of the Hill. QIC Tape and Its Descendents. Travan. Scalable Linear Recording (SLR). 4 mm and 8 mm Contenders. Digital Data Storage (DDS). Mammoth. Advanced Information Technology (AIT). New Players Enter the Mid-Range. Linear Tape Open (LTO). Ecrix and VXA. Into the Rack. Back to NFR. Endnotes.

9. Wither Optical Technologies?

Magneto-Optical Technology Evolves. LIMDOW. Phase-Change Optical Media. Media Selection Criteria. DVD Formats Proliferate. Seeing the Future of Optical Clearly.

III. STORAGE MANAGEMENT TECHNIQUES.

10. Implementing Enterprise Storage Management.

Needs Assessment Phase. Policy Development Phase. Architecture Definition Phase. Storage Management Tools Evaluation Phase. Implementation and Testing Phase. Monitoring and Management Phase. Change Management Phase. Model versus Reality. Continued Obstacles. Endnotes.

11. The Storage Infrastructure Perspective.

From a Repository to an Infrastructure View. A Dynamic Data Movement Example: HSM. SAN Architecture Forces Storage Infrastructure Viewpoint. Storage Infrastructure Requirements for Data Warehouse Applications. Storage-Related Data Layout Requirements for Databases. Storage Infrastructure Requirements for Multimedia Applications. The Storage Infrastructure Perspective. Endnotes.

12. Back to the Future.

Considerations for Selecting a Backup Strategy. Remote Mirroring. Hierarchical Storage Management. The Impact of HSM on Storage Planning. Endnotes.

IV. CONCLUSION.

13. The Holy Grail.

Sun Microsystems And StoreX. Compaq's ENSA. EMC'S Enterprise Storage Networks. Other Visions. The Quest Continues. Endnotes.

Glossary.
Index.

Preface

PREFACE

Data storage and its management are rapidly becoming key issues for modern corporate information technology. While the job title storage manager may be commonplace in the glass house of the traditional mainframe environment, such a title rarely exists in the realm of distributed computing. In the distributed world, responsibilities for managing storage have been largely distributed with the servers to which storage devices have been attached.

A system administrator, with or without software tools beyond those provided as part of the operating system, is typically responsible for managing server storage. He or she provides storage planning, grooming, and other management tasks as a subset of general system maintenance effort.

This arrangement may have proven adequate as long as distributed servers were not tasked with hosting mission-critical applications. However, with the evolution of client/server computing as a trusted platform for mission-critical work, the emergence of Storage Area Networks (SANs) and other nontraditional storage configurations, and the new storage-centric focus of enterprise IT, such casual handling of storage management responsibilities is increasingly less appropriate.

Case in point: A number of SAN product vendors have observed that product sales have been somewhat impeded by the job skills requirements inherent in SAN management. The job title most involved in SAN acquisition decisions is that of system administrator, rather than network manager. Understanding and managing SANs (as well as their cousin technologies such as Network Attached Storage NAS devices) requires a blending of skills from the network, system, and storage management disciplines. In the absence of personnel possessed of a such "hybrid" skills sets, most SAN sales have been driven by demonstrations of specific applications, such as faster backups, rather than by presentations of the design and capabilities of the technology itself.1

In short, the acquisition of newer storage technologies such as SANs is being delegated to nonnetwork savvy system administrators, many of whom do not even possess the knowledge, skills, or tools required to manage even traditional server-captive storage configurations effectively. It is the argument of this book that the situation reflects a potential disaster inherent in unmanaged storage within the corporate IT enterprise.

PURPOSE OF THIS BOOK

The purpose of this book is to provide a primer of sorts that will enable readers to begin cultivating the hybrid knowledge and skills they require to serve as storage managers within the modern IT enterprise. The scope of the book is large and its chapters address a broad range of subjects—from the particulars of storage devices themselves, to the fundamentals of interface and interconnect technologies, to the principles of effective management, to industry "best practices" in planning and analysis.

The book also endeavors to present the current "politics" behind storage technologies. Storage technologies do not develop in an apolitical vacuum, but in a competitive market where vendors vie constantly for a share of the customer base.

Generally speaking, vendor attitudes manifest more a kammeralist than a merchantilist bent. This means that vendors tend to perceive markets for their goods as fixed in size and dollar volume. Sales is a "zero sum game" in which my success is your loss. Only in cases of new technologies that create entirely new markets does the merchantilist worldview of expanding markets prevail.

Against this backdrop, the language of marketing, rather than technology, tends to obfuscate efforts to understand what products are appropriate for a given set of requirements. To the extent possible, this book endeavors to acquaint readers with the code words and buzz phrases of storage "market speak" in order to assist them in separating the kernels of practical information from the chaff of hyperbole and spin doctoring that often accompanies marketing communications.

It is expected that some material will be familiar to the reader, while other content will be new. This reflects the specific background and experience that the reader brings to the text.

Readers of this book are expected to have either a systems-specific or a networking-specific background: Enterprise storage management requires a mixture of skills from both disciplines.

This book is based on extensive research in the evolving area of enterprise storage technology and includes information taken from hundreds of hours of interviews with vendors and consumers of storage products. It is profusely illustrated with line drawings, product photos, charts, and graphs to aid in communicating complex concepts simply and to facilitate understanding.

ORGANIZATION OF THE BOOK

The book is divided into three main parts. Part One provides a context for thinking about storage management and introduces some of the central themes explored in greater detail in subsequent parts of the book. Also provided in this section are several "generic" distributed storage models. These models are intended for use by the reader as a starting point in conceptualizing, planning, integrating, and managing storage capabilities in a distributed environment.

Part Two provides a detailed discussion of storage technologies themselves. The perpetuation of magnetic disk-based storage is considered at length. Beginning with an evolutionary overview of the technology underlying the magnetic hard disk, the latest engineering advancements in hard disk media, controller, and interface technology are examined.

From the micro-level focus on hard disk technology, we shift to a macro-level view and consider disk storage subsystems. Array technologies based on open and proprietary RAID implementations are considered in detail.

By the late 1990s, these array architectures comprised the preponderance of "tethered" (server-attached) storage subsystems. According to analysts, these arrays will be detached from servers over the next decade and migrated to storage area networks (SANs). How this is being accomplished and what capabilities SAN switches and hubs add to the enterprise storage platform will be considered in detail.

This section also includes a detailed examination of another, evolving, disk-based, storage architecture: network-attached storage (NAS). The untethering of storage and its related I/O operations and wait states from the server supports an "appliance view" of storage in an distributed computing context. We will conclude our examination of magnetic disk-based storage by looking at the "thin-server storage appliance" concept, initial products, and the problems that this approach solves (and creates) for IT planners.

In addition to magnetic disk and disk-based arrays, enterprise storage also includes tape and optical storage technologies. Technologies for near-online storage — including tape and optical storage subsystems — continue to have valuable roles to play in enterprise storage platforms. Their traditional missions in disaster recovery and system backup will be evaluated, together with evolving roles in production applications.

Part Three of the book is entitled Storage Management Techniques. Far from being an exhaustive manual for managing the many varieties of storage that are likely to be deployed within the corporate IT enterprise, this section seeks to provide the reader with a framework for analyzing and assessing the relevance of management approaches.

The section begins with an overview of a "project" whose objective is the implementation of an effective storage management capability. While highly simplified, the project model set forth introduces the generic objectives and tasks that need to be considered as IT professionals endeavor to take control of storage and its management within the corporate enterprise.

Core to defining effective storage solutions is the cultivation of a "storage infrastructure perspective." Looking at storage requirements from such a perspective helps to define the application-specific data layout and data movement requirements that a storage solution must support. This concept is illustrated using examples that range from very large databases (VLDBs), which undergird most of today's mission-critical client/server applications, to large-scale data streaming applications, such as digital video and audio editing, which are the harbingers of the next wave of "killer multimedia applications."

Some traditional applications and data movements persist, including hierarchical storage management (HSM) and backup/restore, which are discussed in detail as the third section concludes. Traditional HSM was a technique for optimizing storage capacity by migrating less-used data from "hot," "online," disk-based media to "near-online" (automated tape library or optical library media) or "off-line" (removable or archival tape) storage. This approach failed to catch on in the distributed environment for a number of reasons, including the falling price of disk storage, the distribution of data and lack of network infrastructure to support large data movements over production networks, and the desire of companies to keep all data online all of the time, particularly for applications such as data warehousing. As with traditional backup/restore applications, HSM is currently experiencing a renaissance of interest within the corporate environment, especially as the effects of unmanaged storage, especially increased downtime, begin to be felt. This chapter looks at the impact of new technologies for remote mirroring and the use of SANs to recast these traditional applications for modern service.

The conclusion of this book has two parts. First, it examines the major, competing initiatives in the storage industry to realize a goal of a "storage nirvana" in which all storage is easily managed, scaled, and shared to meet application and end user requirements. Each of these contending approaches, and the many others that will likely follow, reflect a vendor-sponsor's preferences and product orientation. At present, vendors show little inclination to reach an accommodation that will deliver an "open" approach, appropriate to all companies.

Despite the failure of vendors to deliver a common storage infrastructure, this book concludes that the pace of advancement of storage technology continues to stay ahead of the "storage pain curve" for IT organizations. As the market has demanded more storage, faster storage, and greater storage capacity, vendors have been ready with products to meet the need. While this may be regarded as a good thing, it has also supported the cultivation of a laissez-faire attitude among IT professionals based largely on an unstated belief that the situation will continue indefinitely.

This view, however, is a fallacy. As the unplanned deployment of storage capabilities continues, IT organizations are creating an environment that is prone both to dramatic increases in total cost of ownership and to disastrous interruptions of mission-critical business applications.

Inevitably, companies will be forced to begin a quest for the "Holy Grail" of rational storage resource planning. It is better that proactive steps be taken today than expensive and reactive ones tomorrow. The first step is education.

A Living Appendix

Even as this book goes to press, changes are happening in storage-related technology. To keep readers apprised of these changes, a site has been created on the World Wide Web to serve as a "living appendix" to the book. Visit the site at http://www.stormgt.org to remain current with the latest trends, products, and thinking about storage technology and storage management.

A FINAL NOTE TO READERS

A book of this type is sure to draw criticisms as well as (hopefully) some applause. Advocates of competing technologies have a tendency to engage in "religious" wars over the merits of their preferred vision, approach, or protocol.

Some are more subtle than others. For example, in a recent interview with a Microsoft product manager, the author was advised that there are no problems of heterogeneous file system access if everyone simply abandons UNIX servers and moves to a completely Microsoft-standard server environment. The same sort of message has been received by representatives of Sun Microsystems, vendor of several popular UNIX platforms.

The author harbors no particular preference for one technology over another. It is the position of this book, based on hundreds of hours of interviews conducted with industry experts, end users, and analysts, that heterogeneous computing will continue to be a fact of life for at least as long as this book is likely to remain in print.

One bias that the author does embrace is a sort of real politik when it comes to enterprise IT. As in Hobbes' state of nature, life in an unmanaged IT environment is nasty, brutish, and short. Readers are asked to set to one side any idealistic views they may have of a perfect IT environment where self-directing and responsible end users, IT professionals, and vendors work together harmoniously to realize business goals. Real life suggests that management, while imperfect, is the best hedge we have against an unpredictable and chaotic future.

The Holy Grail, referenced in the title of this book, symbolizes a vision or quest that can never be fully actualized or concluded. This is not to say that IT professionals must engage in a pointless struggle to accomplish what can never be accomplished. Such an existential viewpoint would also make this book a pointless undertaking.

Rather, storage management, like so many other aspects of business technology management, seeks to anticipate what is within our ability to anticipate, so it can be controlled. Secondly, storage management seeks to minimize the harmful consequences of events that are beyond our ability to anticipate.

Given the pace of change in business and in technology, management must flexible and open to new opportunities. Effective management is constantly reinventing itself and redefining its objectives and strategies. It is in this context that the symbol of the Holy Grail is invoked. A well-managed enterprise storage platform is a worthwhile objective that requires constant attention and proactive effort on the part of skilled personnel. It is not, and can never be, a task that is completed once and for all.

ENDNOTES

1.Interview with Bill Lozoff, Director of Marketing, Gadzoox Networks, Inc., San Jose, CA. 1998.



Updates

Submit Errata

