Post on 23-Dec-2015
Linux: Hype, Hope, or Strategic Direction
Educause Southeast Regional Conference
June 29, 2001
Jim Bostick
Virginia Commonwealth University
Linux: Hype, Hope, or Strategic Direction
Examples of hype, hope, and strategic direction Description of Gartner Group’s Hype Cycle Case study – VCU 1998 - present
– Problem at VCU in 1998– How Linux became a solution– Outcomes of Linux selection– Prospects for the future
Relevance of the case study to other situations
Hype – No doubt!
Hope – For some…
“Linux is real and Linux is ready for real business.”– Sam Palmisano, IBM President and COO, Linux World,
January, 2001
“Linux is many, many years away from being an enterprise-ready operating system that can compete with, and challenge, the Windows platform”
– Doug Miller, Group Product Manager, Microsoft Windows Server Group, interview with eWEEK before heading to Linux World, January, 2001
Strategic Direction – For others…
Google Burlington Coat Factory, Auto Zone Lawson, Inc – Japanese convenience stores Hess Oil, Conoco Oil, Royal Dutch Shell Boeing
Which describes Linux?
Hype? Hope? Strategic Direction?
Gartner’s Hype Cycle
What to do?
How do you minimize the risk of adopting a highly hyped but unproven technology without ignoring it completely?
The Problem at VCU in 1998
Demands for computing resources were outstripping capacity – Web-based course materials– Computationally-intensive research– Growing web demands
Limited equipment funds severely challenged the ability to meet each of these needs
First Step
Dramatic and rapid growth in Web Course in a Box overloaded aging SGI IRIX server
Options:– Replace with another SGI or comparable Unix
server at cost of ~$20,000– Replace with Linux server at cost of ~$5,000
No choice but Linux
Early Success
Dramatic performance improvement Zero problems over next 6 months
Next Move
Planned replacement of 1 major web server, another SGI Irix server
Success with Web Course in a Box server gave us confidence to select Linux
Dramatic performance success Dramatic cost success
Major Changes
Web rapidly growing Plan to merge separate campus web servers
while replacing 2nd major web server, an IBM RS6000 AIX server
Budget did not limit options Committed to all Linux for web servers
Decision Time
Demand for computationally intensive research overwhelmed SGI Origin 2000 server
Plan to merge separate campus research servers
Options: Add 8 processors to SGI at cost of ~$110,000 Purchase 32 processor Linux Beowulf cluster at cost of
~$70,000
Decision Time
Cost advantage clear, but– Who was using Beowulfs and how successfully?
NIH, NSF in numerous locations U. of New Mexico, Ohio Supercomputing Center, NYU Medical
School, Princeton, U.S. Air Force Academy– What applications run well?
Coarsely parallel applications using MPI and PVM – Any comparison data versus more traditional platforms?
Amerada Hess Oil U.S. Air Force Academy
– Does VCU have applications that will run well on Beowulf? 4 of 5 most CPU intensive applications should run well
Hard Sell
Accused of having been sold on the hype Facts persuaded
– Low risk, high value now Proven success on VCU applications Dramatic cost benefit
– Potential major long term benefits
Outcomes
Linux Web successes– superior performance and cost savings met needs and left
room for growth
Linux Beowulf successes– Comparable, sometimes superior performance– Met current needs with room for growth
Linux Beowulf problems– Implementation slow– Adoption slow
Prospects for Future
In position to adopt Linux for additional needs if applications continue to develop
– Web Implementing Blackboard v5 w/ Oracle Implementing Java Application Server
– Research Computing Bioinformatics Visualization? Statistical analysis? Database apps?
– LAN servers -- ?– Desktop -- ?
Cost savings will continue to be dramatic
Relevance Beyond Linux?
New technologies can provide great benefits… …but with great risks Always will be pushed to implement the newest
hyped technology Generalizing this case study gives a framework for
– deciding which new hyped technologies to buy into – minimizing your risk– positioning yourself to take advantage of the technology
should the hype prove real
The Framework
Decide which new technologies to implement– Carefully evaluate the immediate usefulness or benefit of
the technology in the organization before adoption Minimize your risk
– Implement the technology where there is a need or benefit now
– If there are no further benefits, eliminate or don’t grow the technology
Position yourself to take advantage of the technology should the hype prove real
– Experience and expertise gained allows you to extend use of the technology rapidly
Questions or Comments?
Jim Bostick
Virginia Commonwealth University
Jim.Bostick@VCU.EDU