 |
|
DAILY NEWS AND INFORMATION
FOR THE GLOBAL GRID COMMUNITY /
|
Special Features:
7 STEPS TO SUCCESSFUL GRID
COMPUTING
By Brooklin J. Gore, Senior Fellow, Micron Technology
Inc
This report along with two other new, groundbreaking Grid reports from IDC
and the Economic Strategy Institute will be released at Gt'04 and included
with conference registration.
www.gridtoday.com/04/conference/index.html
This is the first installment of a three-part series derived from the report
"Enterprise Grids For General Purpose Computing" by Brooklin J. Gore.
Introduction
Most folks these days have heard, at least in passing, of Grid (or
Distributed) Computing. Historically, Grid Computing has been associated with
solving very specific, generally large, complex problems. The SETI@Home
(Search for Extra-Terrestrial Intelligence) project is likely the best known
for its use of millions of Internet-connected PCs via a clever screen-saver.
The success of SETI@Home encouraged other Grid applications such as the
(protein) Folding@Home project, which investigates potential cures for
Alzheimer's, AIDS and Parkinson's, and Butterfly.net, a Grid-based gaming
network. The latest "personal" Grid Computing project is
ClimatePrediction.net, which hopes to facilitate better understanding of the
probability and impacts of global warming. Grid Computing has also enjoyed
early success in academic and commercial organizations. However, these
"enterprise" Grid applications have been focused on a narrow set of problems
such as gene sequencing, numerical analysis, simulation and modeling. In most
cases, Grid applications are written by software developers specifically for
Grid execution.
Web services, utility computing, .NET, CPU harvesting and distributed
computing are just a few of the technologies that fall under the Grid
computing umbrella. Gt04 -- a premiere enterprise Grid computing conference
targeting industrial and commercial users -- will gather experts, and outline
strategies and road maps for Grid deployment. For more information, visit
www.gt04.com.
Grid computing is here!
Experience at a global manufacturing company over the past two years has shown
that today, general purpose Grid computing is a reality. What's more, it isn't
difficult, is easily grasped by in-house information technologists and
effectively runs not only "home-grown" software, but also third party
applications that were not developed with the Grid in mind. This article
suggests seven key practices for successful general purpose Grid Computing,
summarizes 10 different general purpose Grid applications and concludes with
challenges and next steps. A Grid Primer is provided in the gray box below.
Now, let us look at what it takes to deploy a successful Grid and more
importantly how to get folks to use it.
Seven Keys To Successful Grid Computing
Just because the power grid can provide standardized, ubiquitous power, it is
not guaranteed everywhere if the local facility can't support it. Anyone that
has lived in a 100-plus-year-old home will attest to that. First, you will
need lots of three-prong adaptors. There is nothing more frustrating than
turning on your new 500 gigawatt home entertainment center and immediately
blowing a fuse or having to return that new arc welder because the dryer is
already plugged into the only 220v outlet and for some crazy reason your
spouse thinks that dry clothes are more important than metal yard art. So it
goes with Grid Computing. There are three key infrastructure requirements for
a pleasant Grid experience:
A solid network. If you are still running a 10 Mb shared Ethernet network,
it is time to upgrade. If you have a 100 Mb switched network, you are good to
go. Gigabit server connections? Even better. Grid Computing will place
additional burdens on your network. Imagine 100 or 1,000 jobs all trying to
get data from a single server -- the network load at the server could be quite
high. Accessible data. Speaking of data servers, they should be accessible from
as many machines as possible. Databases are pretty easy since clients are
generally available for every platform. File servers can be a little trickier
as they historically have fallen into two camps: one for Unix systems (ie:
NFS) and another for Windows systems. There are some solutions available today
that provide access to both platforms in a single server. If your data is
scattered around on individual computers and local hard drives, you should
seriously consider centralizing it with file servers. Standard system images. It doesn't matter if you have IBM mainframes,
Solaris servers, Linux workstations and Windows desktops because most Grid
software automatically knows what platform a job needs and gets it there.
However, for each platform it is important that all systems are as similar as
possible. While most Grid software can determine the architecture, operating
system, memory size, etc. of a machine, it is more difficult for it to know
that Oracle 12.5 drivers are installed. it is best to ensure that all systems
of a given platform have the same applications, drivers, libraries, etc. of
compatible versions.
If your infrastructure has a solid, high-speed network with plenty of
headroom, centralized databases and file servers and standard OS and
application builds for all platforms you are off to a great start. If not,
take two aspirin and read the rest of this article in six months when your
infrastructure is up to par. Once your infrastructure can support a successful
Grid deployment, you can't just use a "build it and they will come" approach.
Below are four keys to ensuring enthusiastic, general purpose use of your
Grid. Articulate the value proposition. Human nature resists change and Grid
computing is a radical change in how we are used to writing and running
programs. "What do you mean I won't know and shouldn't care where my program
is running?" is a typical developer question. Grid Computing is not hard, but
it is different. Use familiar analogies. For example, submitting a job for the
Grid is similar to running a job via cron (a Unix scheduler) or Windows NT
Scheduled Tasks. Highlight the time savers -- not worrying about hardware
failures and performance problems is a big load off a developer. Describe new
capabilities -- the ability to complete a task in a tenth of the normal time
is a sure way to win kudos from the boss (and when you tell him it did not
cost the company anything you might just get that promotion you've been
waiting for). Focus on low-hanging fruit. Do not go after the holy grail at first. It
will take too long, you will get frustrated and others will note your
struggles and lose interest. Find some type-A adopter or sponsor in your
organization with a simple, but important problem that is well suited for the
Grid. Slowly but steadily, demonstrate the successes and benefits of Grid
Computing. Many small but solid wins time after time will get more attention
than the single hail Mary that took two years. Educate, educate, educate. As a Grid pioneer, it is your responsibility to
ensure success, not everyone else's just because you say so and it is sexy
technology. Educate everyone, early and often. Educate management on the cost
savings and competitive advantages Grid Computing offers. Educate developers
on the approach, benefits and ease in deploying Grid applications. Educate the
infrastructure support staff on the ease of managing a Grid environment. When
the light comes on in other folks' heads and they have great ideas for the
Grid, you will know your approach is successful. Integrate with existing models. Do not re-invent the wheel or create new
organizations or processes. Grid Computing is not so radically different to
require organizational or procedural changes. You do not need a new
development team or a specialized Grid support team -- you will be challenged
by management and frustrate anyone left in the "old" organization. It is much
better, through education, to simply integrate Grid Computing into existing
development and support policies and procedures.
Now, let us see what can be done with Grid Computing when these key practices
are used.
About Brooklin J. Gore
Brooklin Gore has been researching and implementing enterprise Grid
technologies for the past three years to create Micron's global Grid
infrastructure which runs over 15 production applications today. Brooklin has
been with Micron for 16 years. In that time he served as a product engineer,
Computer Aided Design group manager, network manager and general manager of
Micron's Internet Services Division. Brooklin has been issued several U.S.
patents and is a senior member of the IEEE. He holds Bachelor of Science
degrees in Computer Science and Electrical Engineering from the University of
Idaho and a Masters of Science in Computer Science from the National
Technological University.
|