Windows NT at D0

Paper: 353
Session: G (talk)
Speaker: Fuess, Stuart, Fermilab, Batavia
Keywords: off-the-shelf products, commodity computing, PC operating systems, desktop computing


Windows NT at D0

Stuart Fuess
Fermi National Accelerator Laboratory
PO Box 500
Batavia, IL 60510 USA
D0 Experiment

Abstract

D0 is currently in the process of replacing nearly 200 VAX
workstations in preparation for activities related to Run II
of the Fermilab Collider. Most of the VAX/VMS nodes provided
desktop environments for users. D0 is actively pursuing
the use of PCs running Windows NT as the principal new
desktop machine. This paper will discuss D0's usage of NT
and our experiences to date.

The D0 computing philosophy for the Run II era is one with
a flexible hierarchy of machines. We expect that bulk data
access and processing will take place on large central UNIX
platforms with access to Terabytes of disk and many Terabytes
of robotic storage. The next level in the hierarchy will be
several workgroup servers, UNIX or possibly NT, with access to
reduced data sets by groups of users. Another level will be
occupied by UNIX workstations, active in code development and
dedicated small data set processing. The final level is the
desktop, which we expect to be populated principally with PCs
and X-terminals.The goals for the desktop are inexpensive access to all other
levels of computing, with additional easy access to common
desktop 'office' applications. It is anticipated that some
number of desktop machines will also be used for code
development and/or local analysis. D0 has chosen to pursue
Windows NT as the operating system for these nodes.
Additionally, X-terminal access to the NT applications is
provided by a server running the WinCenter product.

This paper will describe the the configuration of our current
system of approximately forty NT platforms, detailing the
hardware and software products in use. We will note
operational modes and experiences, with concentration on how
both scientific and technical users have adapted to the
environment. Finally, we will comment on the surprises we
encountered in the deployment of these systems.