Monday, 11 August 2014

Designing Storage Area Networks: A Practical Reference for Implementing Storage Area Networks

feedback Feedback | Switch to single page view (no tabs)
From the Book:
The following work expands and updates the content of my previous book, Designing Storage Area Networks: A Practical Reference for Implementing Fibre Channel SANs, published in the winter of 1999. At a thin 202 pages, that book was the first to provide a brief overview of storage area network (SAN) technology and was widely adopted as an introductory and training text for both vendors and customers. Due to the rapid growth and technical development within the storage industry over the past three years, however, new functionality and technology initiatives are already transforming the SAN landscape. Advances in Fibre Channel performance and switch technology, the introduction of SANs based on TCP/IP and Gigabit Ethernet, and the emergence of storage virtualization have given customers more options for addressing their data storage needs. While some of these technical initiatives are presented in my second book, IP SANs, A Guide to iSCSI, iFCP and FCIP Protocols for Storage Area Networks, this text provides more focus on the practical considerations implementers should consider in designing SAN solutions with today's diverse technologies.
Storage area networks are now recognized as the preferred solution for fulfilling a wide range of critical data storage needs for institutions and enterprises. The success of SANs in establishing real end-user value is affirmed by the adoption of the technology by all major solution providers as their flagship server and storage offerings. IBM, Hewlett-Packard, Sun, Dell and others now provide certified SAN configurations for performance, high availability and backup of storage data, while storage vendors such as EMC,Hitachi Data Systems, XIOtech and others offer SAN interfaces on their premier products. While the current market penetration of SAN-based solutions is only 20% of the total storage market, SANs are expected to capture the majority of the market within a few years.
The emergence of storage networking represents the fusion of two distinct technologies, each with its own priorities, vocabulary and culture. For twenty years, the glacial pace of storage development gradually resulted in smaller, higher capacity disks and faster channel performance, but this evolutionary inertia was always bounded by a rigid master/slave relationship between hosts and storage. By contrast, the rapid and more volatile development of networking technology has overturned previous architectures and relationships and, as demonstrated by the Internet, has delivered a global capability with innovative and enriched feature sets. Combining the more conservative outlook of storage with the more dynamic worldview of networking has resulted in a collision of cultures that has actually benefited both disciplines. Data storage has been transformed from the frumpy wallflower of data communications into a stellar attraction for investors, vendors and technologists. Networking has gained an endless supply of high volume data that further extends and justifies its presence throughout the enterprise, as well as new opportunities to create storage-specific interfaces and functionality.
This synthesis of storage and networking into a new technology has presented both challenges and opportunities for customers and vendors. Storage administrators and managers must now be conversant in networking concepts and consequences such as addressing, routing and network convergence, while network architects and administrators must learn the mysteries of LUNs, JBODs and RAID levels. The following work is therefore written for a broad audience of IT managers, administrators, consultants and technical staff whose responsibilities may span both storage and networking implementations. In addition, storage networking has generated new concepts and issues previously undefined by either storage or networking. The unexpected consequences of positioning storage in an open, networked environment are also examined in the text and particularly in the applications studies.
While storage networking is an enabling technology for dealing with the massive growth of storage data, it is often criticized for persistent interoperability and management issues. Some of the problems associated with SANs are simply the byproduct of a new technology struggling to break new ground. Some of the problems are, unfortunately, vendor-induced in an effort to retain market share. Overcoming interoperability issues and management complexity are prerequisites to wider-scale adoption of SAN technology, and several industry initiatives are now attempting to address the remaining obstacles. Although some of these efforts may take several years to bear fruit, they are discussed in the following text to indicate where help is on the way.
In the first edition of Designing Storage Area Networks the underlying infrastructure or plumbing for SANs was exclusively Fibre Channel. Fibre Channel was the first successful gigabit serial transport and pioneered the signaling and data encoding mechanisms later adopted by Gigabit Ethernet. Had Gigabit Ethernet appeared first, it might have preempted Fibre Channel as a storage network transport. While Ethernet has on its side the massive momentum of market presence in the rest of the data communications space, its tardy arrival in storage networking now puts it in conflict with the flourishing Fibre Channel. Although this contention has generated ongoing religious disputes among vendors, both Fibre Channel and Ethernet are, after all, simply plumbing. The following text provides technical detail on both Fibre Channel and Ethernet-based SAN technologies in the expectation that customers will make their own decisions on which transport best meets their needs. Today, the vast majority of vendors selling SAN solutions are planning to provide both Fibre Channel and IP-based products to accommodate their customers' requirements.
The people who have direct responsibility for designing and implementing SANs desperately need more detailed information on product capabilities and interoperability, i.e., what version of a particular backup application works with a specific database or operating system level. Given the lead time required for hard copy publishing, it is impossible to capture this very useful but granular information. By the time such a text hit the stands, it would be outdated and of little value. The Storage Networking Industry Association (SNIA) is currently organizing end-user organizations with online databases of customer issues and solutions. I urge any reader seeking more detailed information on SAN solutions to become involved in this SNIA-led effort and help build the practical resources that will benefit the user community at large.
Designing a storage network requires, first of all, answering the basic question: What is the application? Customers do not, after all, spend millions of dollars on storage arrays, tape subsystems, switches, servers and cabling for the sake of the technology itself. Applications are fostering SAN deployment, and applications are as varied as the core requirements of diverse institutions and enterprises. Although vendors may package canned solutions to fulfill common needs such as storage consolidation or tape backup, the SAN architect should avoid being driven by vendors but instead drive the vendors to solve the customer's specific application requirements. This text attempts to define the most common applications that are more efficiently served by SAN technology, but these examples only begin to address a much broader spectrum of real-world applications that storage networking may benefit.
OrganizationThe present work will take the reader through a progression of concepts that in the end should provide a foundation for understanding storage networking infrastructures and applications. It assumes no prior knowledge of SANs and yet attempts to strike a balance between basic and advanced content. This will hopefully provide sufficient technical detail for those who need it and meaningful overview for those who want to understand SAN technology at a more abstract level.
The first two chapters provide a framework for understanding the central concepts of shared storage. Chapter 1 reviews the Storage Networking Industry Association (SNIA) Shared Storage Model, which clearly abstracts the basic layers of storage applications and underlying infrastructure. The SNIA Shared Storage Model is a useful tool for understanding SAN architectures and positioning them in relation to upper layer application requirements. It has also proven to be a useful tool for justifying SAN acquisitions to upper management by explaining how a particular solution will better service business requirements. Chapter 2 provides an overview of storage and networking concepts and explains how the fusion of these ideas has created new means to solve data storage issues.
Since Fibre Channel was the first transport to appear on the storage networking scene and continues to ship in significant volumes, the next three chapters provide a technical discussion of Fibre Channel protocols, topologies and products. Chapter 3 reviews lower layer physical transport, protocols and addressing. Chapter 4 reviews Fibre Channel topologies, with more emphasis on the Fibre Channel fabric switch services provided by most SAN solutions today. Chapter 5 describes the product suite vendors have developed for Fibre Channel servers, storage, interconnects and other components. Collectively, they provide a rich toolset for the SAN architect in designing efficient solutions.
Beginning in 2000, the initiative to transport storage data over mainstream TCP/IP networks generated IP storage networking architectures and products. These products further enhance the ability of SAN designers to craft shared storage configurations that leverage IP exclusively or integrate both IP and Fibre Channel into a heterogeneous storage network. Chapter 6 reviews the protocols and issues unique to IP SANs, while Chapter 7 describes IP storage products that have been introduced to the market over the past two years.
Chapter 8 provides an overview of storage-specific applications supported by the SAN infrastructure. Server clustering for high availability data access, tape backup for data archiving, and data replication via disk mirroring are commonly deployed as part of a SAN solution, although these storage applications in turn are intended to service high level customer business applications and databases.
Inevitably, the gains that SANs provide are accompanied by some pain. While vendors attempt to debug and certify SAN configurations in advance, problems naturally arise during installation or subsequent operation. Chapter 9 explains some basic troubleshooting techniques and tools available to identify and isolate storage network problems.
Chapter 10 discusses SAN management and its unique challenges. Management of shared storage requires the integration of transport management with management of data placement on disk or tape. Creating a unified management capability has been difficult, even for software vendors with ample resources to address SAN management. The Common Information Model (CIM) initiative, however, is at last getting traction in the industry and may provide the framework required for comprehensive management of storage networks.
Chapter 11 attempts to describe the amorphous entity known as "storage virtualization." Storage virtualization has suffered somewhat from excessive vendor marketing, but is a very viable technology for simplifying storage administration and opening a broader market for SAN solutions. This chapter covers the theoretical capabilities of storage virtualization as well as the more limited functions currently available in shipping products.
Institutions and enterprises have been implementing SANs for a wide variety of applications. Some of these are discussed in Chapter 12, Application Studies, with examples of potential issues customers might face in the course of deploying and administering SANs. Where applicable, combinations of Fibre Channel and IP storage products are proposed, especially when addressing metropolitan and wide area SAN configurations.
Chapter 13 discusses the still unresolved issues of SAN technologies, including interoperability, management and convergence between SANs and mainstream networking. For engineers, these issues are simply challenges to be mastered, providing market competition does not artificially thwart their efforts.
As a summary of SAN design, Chapter 14 offers speculation on the future of storage networking. That future promises to be as dynamic as the emergence of SANs was, with the rapid development of new storage initiatives such as virtualization, IP-based SANs and accommodation of new host interconnects such as InfiniBand.
In this edition I have included in the Appendices additional material that may be useful as reference or background information on storage networking in general. The Storage Networking Industry Association (SNIA) has produced some very valuable work in the form of the SNIA Shared Storage Model and the SNIA Dictionary. I encourage both vendors and consumers of storage networking technology to take advantage of the SNIA's ongoing work and, when possible, to participate in SNIA's activities. The more customers and technologists align their common goals, the more quickly viable solutions can be put into the customers' hands.

Powered by The ACM Guide to Computing Literature

The ACM Digital Library is published by the Association for Computing Machinery. Copyright © 2014 ACM, Inc.
Terms of Usage   Privacy Policy   Code of Ethics   Contact Us

Useful downloads:  Adobe Reader     QuickTime     Windows Media Player     Real Player