Archives
 
 
 
  Special
 
 
 
  About Us
 
 
 

Newsletter
Free E-mail Newsletter from BYTE.com

 
    
           
Visit the home page Browse the four-year online archive Download platform-neutral CPU/FPU benchmarks Find information for advertisers, authors, vendors, subscribers Request free information on products written about or advertised in BYTE Submit a press release, or scan recent announcements Talk with BYTE's staff and readers about products and technologies

ArticlesThe Server that Wouldn't Die


October 1997 / Reseller / The Server that Wouldn't Die

High availability and fault tolerance are coming to servers near you.

David Baum and Gregory Karpain

When your power goes out, it's upsetting. If your phones go down, it's catastrophic. So why are we willing to accept less from our computer systems? Simple: The power company and the phone company have money. Lots and lots of money.

Traditional high-availability (HA) solutions are very expensive, limiting their customer base to the likes of Fortune 500 companies. However, new, less expensive HA systems are catching the eye of smaller companies. Even fu lly fault-tolerant (FT) systems are coming down in price. Customers are starting to expect HA in their servers.

Why HA Now?

According to Donna Scott of the Gartner Group, the lower costs of HA technology and production, the increasing prevalence of the Internet, and the growth of globalization, mobile computing, and 24-hour, seven-day-a-week customer-service requirements are all fueling the fire for HA solutions.

Where big-league HA solutions used to cost $1 million and up, Unit Systems, an NCR VAR, can now offer top-notch systems ranging from $250,000 to $300,000, says Christopher Radzik, president of Unit Systems. This puts the technology within the grasp of some of today's fastest-growing markets, such as those required by Internet service providers (ISPs) and for companies installing voice-processing systems. Armed with ammunition such as this, companies outside the Fortune 500 are installing HA systems.

Companies and Technologies

Computer vendors have been quick to respond with HA and clustering solutions for the VAR and end-user community. Data General, Digital Equipment, Hewlett-Packard, IBM, NCR, Silicon Graphics, Sun Microsystems, and myriad small er vendors have taken up the HA torch. Tandem Computers and Stratus Computer lead the pack with FT solutions.

For example, IBM has extended its extensive experience with highly available systems in the mainframe arena to its RS/6000 and AS/400 platforms, where its High Availability Cluster Multi-Processing (HACMP) software brings HA solutions down into the midrange turf. Digital has leveraged its pioneering VAXcluster technology into its AlphaServer line with TruCluster software for Unix and OpenVMS systems.

HP is another solid contender, with its HP-9000 midrange servers and MC Service Guard clustering software. NCR, one of the biggest suppliers of HA solutions, offers midrange solutions using its WorldMark 4300 with LifeKeeper clustering software. Sun, a latecomer to the HA field, offers an HA solution on its Sun 4000 and 6000 enterprise servers. When it comes to FT systems, Tandem rules the roost with its NonStop Himalaya systems. Stratus comes in second with its Continuum series.

Buildi ng Experience

Server technology alone does not make an HA solution. "Proper administration and maintenance of clustering systems is one of the biggest challenges in the mid-market today," says Tom Foley , president of Baystate Computer Group. "You can invest $2.5 million in a sophisticated HA solution but can't expect to administer it with a $45,000 technician. The customer must train, train, and train again."

The Gartner Group's Scott suggests that customers build HA "practices" in their companies. "An HA solution is one-third technology, one-third business practice, and one-third organization," she says. "Organizations must develop proper HA procedures; train personnel; automate wherever possible to prevent user error; implement good testing, deployment, and configuration management procedures; and look at network and systems software that predicts problems," Scott summarizes. "Successful HA is not an out-of-the-box solution. It is a complex formula that must be expertly e xecuted."

HA vs. FT

The degree of HA needed for each application depends on the risk factors: How devastating is downtime, and how much is the customer willing to spend to prevent it? In most organizations, 99 percent availability is good enough. Even the phone company promises only "five nines," or 99.999 percent uptime.

What makes the difference between 99 percent and 100 percent availability? Clustering technology exists that can achieve a failover swap in less than a second. But what about the transaction in process? You can roll it back and reconstruct it in a matter of minutes. For many companies, this is acceptable. For others, 100 percent availability means nonstop computing, without even a nanosecond of downtime or a byte of data lost. For these solutions, customers turn to FT technology (see the sidebar "Defining Terms").

Traditionally, FT systems have come with the highest price tags. However, improvements in production and technology similar to those that have benefited HA solutions are making FT solutions more affordable as well. "A business can now afford an FT system at only 30 percent higher cost than an HA-cluster solution," says Jon Howe, who is the chief technical officer at ACI. "As with any technology, we are always pushing the price down over time."

Will the Twain Ever Meet?

As it stands today, HA and FT are two different technologies with two different levels of availability, but most experts agree that the distinction is growing smaller. HA is even cropping up on PCs, as Microsoft comes to market with its clustering solution for Intel-based processors. This will create yet another set of potential markets, driving HA solutions even further down the food chain.

In the near future, we will also see an intermarriage between HA and FT technologies, further blurring the distinction between them. Already, the level and kind of availability that can be created by either HA or FT solutions is approaching a difference of only a few decimal points. Ta ndem and Microsoft are working together to blend the two types of systems, incorporating Tandem's FT nonstop-kernel philosophy -- hide the failure from the application -- with Windows NT's HA-server philosophy -- failover and restart the application.

But no matter how you slice up the market or define the terms, the driving need for HA systems remains the same. As people in all types of businesses come to depend on their computer systems almost as much as the electricity that powers them, they will be less and less tolerant of outages of any kind. "People won't tolerate the lights going out in their offices for an hour every week," Radzik concludes. "Why should they expect any less from their computer systems?"


Where to Find


Digital Equipment Corp.

Maynard, MA
Phone:    800-344-4825
Internet: 
http://www.digital.com


Hewlett-Packard

Palo Alto, CA
Phone:    800-857-1501
Phone:    650-857-1501
Internet: 
http://www.hp.com


IBM

Armonk, NY
Phone:    800-426-4968
Internet: 
http://www.ibm.com


Microsoft Corp.

Redmond, WA
Phone:    800-426-9400
Phone:    425-882-8080
Internet: 
http://www.microsoft.com


NCR

Dayton, OH
Phone:    800-2
62-7782
Phone:    937-297-5700
Internet: 
http://www.ncr.com


Silicon Graphics, Inc.

Mountain View, CA
Phone:    800-800-7441
Internet: 
http://www.sgi.com


Stratus Computer, Inc.

Marlboro, MA
Phone:    508-229-4000
Internet: 
http://www.stratus.com


Sun Microsystems

Mountain View, CA
Phone:    800-555-9786
Phone:    800-786-0404
Internet: 
http://www.sun.com


Tandem Computers

Cupertino, CA
Phone:    800-538-3107
Phone:    800-482-6336
Internet: 
http://www.tandem.com


HotBYTEs
 - information on products covered or advertised in BYTE


Tom Foley

photo_link (29 Kbytes)

"Proper administration and maintenance of clustering systems is one of the biggest challenges in the mid-market today."


David Baum and Gregory Karpain are freelance business writers specializing in information-technology topics. You can contact them by e-mail at dwbaum@silcom.com and gkarpain@silcom.com .

Up to the Reseller section contentsGo to previous article: Go to next article: Defining TermsSearchSend a comment on this articleSubscribe to BYTE or BYTE on CD-ROM  
Flexible C++
Matthew Wilson
My approach to software engineering is far more pragmatic than it is theoretical--and no language better exemplifies this than C++.

more...

BYTE Digest

BYTE Digest editors every month analyze and evaluate the best articles from Information Week, EE Times, Dr. Dobb's Journal, Network Computing, Sys Admin, and dozens of other CMP publications—bringing you critical news and information about wireless communication, computer security, software development, embedded systems, and more!

Find out more

BYTE.com Store

BYTE CD-ROM
NOW, on one CD-ROM, you can instantly access more than 8 years of BYTE.
 
The Best of BYTE Volume 1: Programming Languages
The Best of BYTE
Volume 1: Programming Languages
In this issue of Best of BYTE, we bring together some of the leading programming language designers and implementors...

Copyright © 2005 CMP Media LLC, Privacy Policy, Your California Privacy rights, Terms of Service
Site comments: webmaster@byte.com
SDMG Web Sites: BYTE.com, C/C++ Users Journal, Dr. Dobb's Journal, MSDN Magazine, New Architect, SD Expo, SD Magazine, Sys Admin, The Perl Journal, UnixReview.com, Windows Developer Network