Six Server Segments: Guidelines for Categorizing Servers

Wednesday Jan 5th 2000 by ServerWatch Staff
Share:

In the world of Wintel-based servers, the types offered are as varied as the applications for which they are used. Although applying broad terms is risky, this tutorial breaks the server market into six segments and provides guidelines for the features and characteristics of each.

By Richard Krause and Bradley L. Hecht

Overview

In the world of Wintel-based servers (aka PC servers), the types of machines offered are as varied as the applications for which they are used. Although applying broad terms is risky, we have broken the server market into six segments. Here, we provide a guide to the features and characteristics of each. (There are other ways the market can be segmented — we have chosen this for its "accessibility" to the general user.)

In general, servers can be segmented by the size of the user population they serve — from small "mom and pop" shops up to large companies with thousands of employees. However, there are exceptions, which will also be explored herein.

Segments

The size-related segments we have defined, in order of smallest population served to largest, are:

  1. Workgroup
  2. Department
  3. Midrange
  4. Enterprise
  5. Superenterprise

All of the above segments are for general-purpose servers — those that try to have a broad mix of features and functions to appeal to the widest range of customers. The sixth segment, is relatively new and is only a few years old. Rather than being defined by the number of users, such servers are categorized in terms of what they do, not whom they serve. This segment is variously referred to as "single-function servers," "server appliances," "information appliances," "thin servers," and "task servers" — each designation having slightly different connotations. For simplicity, we will lump the subgenres together and refer to the general product group as "appliance servers." This segment (for both Wintel and Unix/Linux-based systems) is expected to reach $8 billion by 2003, according to IDC.


Richard Krause is Research Director, Hardware Platforms at TechnologyEvaluation.COM. In this role, Mr. Krause manages all of the research and data modeling for computing, storage, and peripheral hardware.

Bradley L. Hecht is executive vice president of TechnologyEvaluation.COM. In that role, Mr. Hecht manages TechnologyEvaluation.COM's global research process and methodology as well as its E-Commerce and Technical Infrastructure practices.

Opinions expressed in this article are those of TechnologyEvaluation.COM and do not necessarily reflect the opinions of ServerWatch or internet.com.

Segment Definitions

For the purposes of this document, we will use the definitions below. Over time, these definitions have changed, and will continue to change. We expect segments will consolidate. Refer to the figure below for an approximate user quantity vs. segment map.



Workgroup:

Users: Generally serves small groups, from as few as as two or three clients or users to up to 20 or 30
Pricing: Usually low-priced (under $2,000)
Typical Uses: E-mail, file, and print serving
Functionality: Medium functionality, positioned slightly above a high-end PC Examples: Compaq ProLiant 800HP, Dell PowerEdge 1300, and NetServer E60
 

Department:

Users: Handles moderate-sized groups, from 10 to 50 or more users (can sometimes handle as many as 100 users)
Pricing: A base price of less than $2,000 for some models
Typical Uses: E-mail, file, and print serving
Functionality: Increased reliability, power, and functionality
Additional Comments: The line between workgroup and department servers is becoming blurred as high-end computing power increases.
Examples: PowerEdge 2300, NetServer LC3, IBM Netfinity 5600, and ProLiant 1600
 

Midrange:

Users: Handles larger groups, starting at around 50 users, can handle groups of up to 100 to 200, or more users
Pricing: Starts at $3,000 to $4,000 for base systems
Typical Uses: Messaging, e-mail, file/print serving, and midrange computer-intense tasks
Functionality: Reliability and performance features become more significant and powerful.
Additional Comments: These servers are sometimes referred to as small to midsize business servers. The line between department and midrange servers is becoming blurred as high-end computing power increases.
Examples: NetServer LH3, Netfinity 5500 M10, ProLiant 3000, and PowerEdge 4300
 

Enterprise:

Users: Handles large groups, 200 to 500 or more users and clients
Pricing: Starting above $5,000, generally in the $7,000 to $8,000 range for a base system; fully configured systems can easily exceed $500,000
Typical Uses: Data warehousing, large database management, and heavy-duty transaction processing
Functionality: Much greater emphasis on reliability, performance, and functionality
Additional Comments: At this level, servers are sometimes architected to have functionality moved "outside the box." For example, hard disk storage within the server may be minimal (just enough to hold the OS), and the server is designed to be rack-mounted with a large RAID array installed in the same rack.
Examples: IBM Netfinity 7000 M10, ProLiant 6500, PowerEdge 6300, and NetServer LH4
 

Superenterprise:

Users: Generally exceeding 500 users/clients, sometimes more than 1000
Pricing: Starts in the $20,000 range for basic systems, but a fully-configured system, plus associated storage, memory, and options can exceed $1 million.
Typical Uses: Data warehousing, massive database management, high intensity transaction processing(>40,000 tpmC)
Functionality: Reliability is extremely important, more so than pricing; computing performance and functionality are also key
Examples: ProLiant 8500, PowerEdge 8450, NetServer LXr 8500, and Netfinity 8500
 

Appliance:

Users: ISPs, small groups
Pricing: About $1,000 for a base system
Typical Uses: Web caching, Web serving, intranet serving, and storage area networks
Functionality: Easy to install and set up, scalability, and CPU density (in rackmount systems)
Examples: Cobalt Networks RaQ2, Whistle Communications InterJet, Netfinity 4000R, NetServer LPr, and Compaq TaskSmart

Correlation Factors

Now that we have defined the segments, we can address the questions of: What can we expect to get for our money, and what is the correlation between market segment and feature/function set? In other words, on which areas (e.g., processing power, reliability, or price) can we expect each of the segments to focus?

The features and functions that matter in servers are separated into two general groups: "goals" (which are more strategic in nature naturedures" (which are more tactical in nature). Goals are the overarching categories, and features are the implementation methods or physical constructs by which the goals are achieved.

The factors listed in Table 1 are a sampling of some of the more important ones; enterprises may find others relevant to their particular situation or need. Different enterprises will have different requirements, and thus factor weightings will vary by enterprise. Because of this, the relative "weight" of each factor is not part of Table 1. Enterprises can use Table 1, in concert with other sizing tools (usually based on the software or infrastructure, such as SAP R/3 or Microsoft Exchange), to help determine which class of server will be best suited to their needs.

Outlook

As the server market consolidates, market/product segment distinctions — especially at the low end — will blur. We believe that within 12-24 months, there will be only three size-related segments (entry/workgroup, department, and enterprise) as well as the general class of appliance servers. As the number of segments decreases, the distinction between them will increase. However, certain patterns will be maintained, such as enterprise servers needing to be extremely reliable. Enterprises should keep these factors in mind as they move toward future server purchases.

Table 1 shows the correlation of the various goals and features to the defined segments.

Table 1
Goals and Features of Each Segment

App. Wkgrp. Dept Midrange Ent. Super-
ent.
Goals
           
Performance/
Power
H
M
M
H
H
H
High
Reliability
L-M
M
M
M-H
H
H
Price
H
H
H
M
M
M
TCO
H
H
H
H
M
M
Flexibility
L
L
M
M
H
H
Service/
Support
L
M
M
M
H
H
 
Features
Multiple
CPUs (qty.)
L
L
L
M
H
H
Storage (Internal)
L
M
H
H
H
M
Redundancy
L
L
M
M
H
H
     Fans
L
L
L
M
H
H
     Power
M
L
M
H
H
H
     NICs
M
L
L
M
H
H
Hot Swap/
Maintainability
L
L-M
M
M
H
H
     Fans
L
L
 M
M
H
H
     Drives
M-H
M
M
H
H
H
     Power
L
L
M
M
H
H
 
How to Interpret
Table 1
Correlation Factor
 L
Low
L-M 
Low/Medium
M
Medium
M-H
Medium/High
H
High
For any given factor, a correlation rating (low, medium, or high) is given for each of the segments. This rating is an indicator of how much emphasis is placed on a particular factor for that particular segment. For example, "price" is very important in the selection of a workgroup or an appliance server, but not quite as important for a superenterprise server. Similarly, "high reliability" is very important in the enterprise and superenterprise market space, but is less of a consideration for an appliance server.

Factor Definitions:

Performance/Power: Computational (or similar) power, the ability to perform a large number of operations or to handle a large workload

High Reliability: The ability to run without a nonrecoverable failure for long periods of time ("reliability" is a goal for all classes of server; the term "high reliability" is used because different server classes require different levels of reliability)

Price: Base system price, or price with a modest complement of features added

Total Cost of Ownership (TCO): An amalgam of fully-configured system price, installation/set-up/training costs, infrastructure costs, downtime costs, and other factors

Flexibility: The ability of the server to be reconfigured for disparate tasks and functionality, as needed by the user.

Service/Support: Related to availability and reliability, maintainability, and ease-of-use considerations.

High CPU Quantity: Whether a server can support four, eight, or more CPUs

Storage(Internal): The storage capacity, both in terms of raw space (i.e., how many gigabytes) and the number of disk drives a server can support inside the chassis

Redundancy: Whether a given component (e.g., fan, power supply, network interface card, or disk drive) has a back-up and a failover mechanism; redundancy is a method for improving system reliability/availability

Hot Swap/Maintainability: The ability to remove and replace a component (e.g., fan, disk drive, or power supply) without having to shut down the system or cause processing functions to stop

Share:
Home
Mobile Site | Full Site
Copyright 2017 © QuinStreet Inc. All Rights Reserved