Read an Excerpt
Configuration and Capacity Planning for Solaris Servers provides information about how to configure aSolaris-based system for use as a server in NFS, database management, Internet/intranet orgeneral-purpose environments. In modern computing, systems are nearly always used in a networkedcontext, so this book considers the impact of a server on its clients and networking infrastructure -and in turn how those components affect the server. This document concentrates on the loadcharacteristics of each type of usage and how this usage interacts with the architecture of variousSolaris-based servers to affect end-user performance.
Although the title might lead one to believe that this book is about Solaris, it is really aboutcomputer architecture - specifically, applied computer architecture. Solaris is certainly a topic, butmore fundamentally the reader should take from this material an understanding of how to apply theprinciples of computer architecture to the solution of everyday configuration and usage problems.
One of the peculiarities of this book is its emphasis on understanding, configuring, and using I/Osubsystems. Virtually every "server" system in the open computing era has an enormous interest inmanaging I/O. Modern server systems certainly have their share of central processing functions, butthis topic is exhaustively covered in the literature and is relatively familiar to many readers. Muchless well understood is the nature of I/O functions, and how to tailor usage, applications, software,and hardware to accommodate the realities of I/O.
This book aims to aid the reader in configuring a serverfor a given end-user requirement, whetherthat requirement is current or future. As such, it is oriented toward people who typically make thiskind of recommendation or who are responsible for managing it once a system is installed: typically,these would be MIS, data processing, or other technical staff. An understanding of the basic conceptsof computer architecture is assumed, as well as intuitive understanding of the requirements ofapplications. The first section of the book, which deals with application profiles, is somewhat lesstechnical, and is suitable for a wider audience. For this reason, the first part of the book makesrecommendations with little or no technical justification. The second part of the book provides thebasis for these recommendations in considerable depth.
How to Use This Book
This document consists of two very different sections. After introductory comments about methodology,the first four chapters - dealing with NFS servers, DBMS servers, Internet/intranet servers, andtimesharing or general-purpose servers - are tutorial in nature. The categories are dealt with inorder of increasing complexity. As a rule, DBMS servers are considerably more complex than NFSservers, and timesharing or general-purpose servers, especially ones with diverse user communities,are the most complex of all. These chapters are intended to be read in their entirety, rather thanbeing used primarily as reference material. Because the topics are quite diverse and may not apply toevery reader, each chapter is written to be independent of the others. Each chapter includes a numberof case studies as well as explicit rules of thumb, along with questions to consider when analyzingreal problems. As noted, in-depth technical explanations are often deferred. The case studies providedirect illustrations of the guidelines at work. Although the examples are designed to berepresentative, they are all completely fictional and do not represent an real site. Any resemblancebetween these case studies and actual sites is purely coincidental and unintentional.
The second section provides detailed technical material on a variety of subjects related toconfiguration planning. Technical bookstores carry a wide variety of excellent books on computerarchitecture, but most of them treat the subject in an abstract manner. The discussion provided inthis text reflects the real-world concerns that are relevant in daily use. This section provides thetechnical justifications for the recommendations and rules of thumb provided in the first section. Itis intended to serve primarily as reference material. There are chapters on each other primarycomponents of architecture: the basic system core (processors, memory hierarchy, and bus structures),storage systems (primarily disk), the backup and recovery process, and the Solaris 2 operating system.For the most part, this section is not specific to Sun products, since much of the architecture is adirect implementation or derivation from well-known industry standards such as SCSI.
Throughout the book, many sections conclude with a group of configuration recommendations thatsummarize the preceding text. These recommendations are marked with the (arrow) symbol.
This book addresses products announced by Sun Microsystems through December 1996. Specificallyincluded are various models of the Ultra-1, Ultra-2, SPARCstation 4, SPARCstation 5, SPARCstation 20,the SPARCserver 1000E, SPARCcenter 2000E and the Ultra Enterprise family, as well as Solaris 2releases up to and including Solaris 2.5.1. Background information is provided about many earlierproducts such as the SPARCstation 2, the SPARCserver 600MP series, and previous releases of Solaris 1,including running in Intel x86 and PowerPC systems is also covered, although the diversity of hardwareplatforms makes this task somewhat difficult.