POWER SOLUTIONS - Dell · Remote Offices Using VERITAS Backup Exec and VERITAS Replication Exec on Dell Hardware By Brian Nelson and Brian Greene 30 Meeting Regulatory Compliance
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
EDITOR’S COMMENTS
6 It’s All About You By Tom Kolnowski
STORAGE
14 Solving the RAID Compatibility Puzzle By Matthew Brisse and Bill Dawkins, Ph.D.
16 Advanced Storage Life-Cycle Techniquesfor Dell/EMC SAN MetaLUN Expansion
26 Efficient, Cost-Effective Data Protection forRemote Offices Using VERITAS Backup Execand VERITAS Replication Exec on Dell Hardware
By Brian Nelson and Brian Greene
30 Meeting Regulatory Compliance Requirementswith VERITAS Enterprise Vault and Microsoft Windows Server Technologies
By Scott Rosen
38 Implementing Cost-Effective RAIDwith CERC SATA 2s Disk Technology
By Harish Jayakumar and Ed Matthews
42 Applying a Modular, Layered Approachto Troubleshooting Tape Drive Hardware
By Gajanan Mudaliar
47 Automating Microsoft SQL Server 2000Online Backup with the EMC SnapView Integration Module
By Ananda Sankaran and Kevin Guinn
MICROSOFT WINDOWS TECHNOLOGY
51 Introducing Microsoft Windows Server 2003x64 Editions for the Intel EM64T Platform
By Ranjith Purush and Chip Webb
56 Deploying Microsoft Windows Server 2003x64 Editions on Dell PowerEdge Servers
By Ranjith Purush and Sandhya Senapathi
62 Guide to Deploying Microsoft WindowsServer 2003 Service Pack 1 on Dell PowerEdge Servers
By Min-John Lee, Scott M. Callaway, and Jeff Ferris
POWERSOLUTIONS
T H E M A G A Z I N E F O R D I R E C T E N T E R P R I S E S O L U T I O N SMAY 2005
COVER STORY | PAGE 8
Advancing Your Enterprise Storage StrategyBy Vicki Van Ausdall
Dell’s scalable enterprise strategy is more flexible thanever. Key storage components now include a budget-conscious iSCSI-based array along with a lineup ofoutstanding CX series Fibre Channel platforms that aredesigned to deliver high availability, performance, andcapacity for demanding enterprise applications.
67 Planning Considerations for MulticoreProcessor Technology
By John Fruehe
73 Achieving High Availability with the Turnkey Dell PowerEdgeCluster FE500W-IA64
By Bryant Vo, Nam Nguyen, and Dat Nguyen
76 Implementing Oracle Database 10g00Maximum Availability Architectureon Dell PowerEdge Servers and Dell/EMCStorage over Wide Area Networks
By Michael Smith, Paul Rad, and Ashish Ray
NETWORK AND COMMUNICATIONS
82 Using Intel Multi-Port Server Adaptersto Enable Virtual Infrastructurein the Data Center
By Bill Henderson and Tracy D. Edwards
SYSTEMS MANAGEMENT
86 Enabling Dell OpenManage Applicationsfor Microsoft Active Directory User Authentication
By Marcos Palacios, Ph.D.
90 Maximizing Remote Management Security on Eighth-Generation Dell PowerEdge Servers
By Chandra S. Mugunda, Weimin Pan, and Haihong Zhuo
95 Advancing KVM Technology Through the Dell 180AS Console Switchand 2160AS Console Switch
By Max A. Benham and Robert Bernstein
THE MAGAZINE FOR DIRECT ENTERPRISE SOLUTIONS
POWERSOLUTIONSWWW.DELL.COM/POWERSOLUTIONS
EDITORIAL
EDITOR-IN-CHIEF | Tom KolnowskiMANAGING EDITOR | Debra McDonaldSENIOR EDITORS | Liza Graffeo, Vicki Van AusdallCONTRIBUTING AUTHORS | Max A. Benham; Robert Bernstein; MatthewBrisse; Michael Brundridge; Scott M. Callaway; Onur Celebioglu; BillDawkins, Ph.D.; Randy DeRoeck; Tracy D. Edwards; Yung-Chin Fang;Jeff Ferris; Ryan Franks; John Fruehe; Brian Greene; Kevin Guinn;Rinku Gupta; Stephen M. Hahn; Landon Hale; Bill Henderson; BruceHolmes; Jenwei Hsieh, Ph.D.; Saeed Iqbal, Ph.D.; Dave Jaffe, Ph.D.;Harish Jayakumar; Monica Kashyap; Garima Kochhar; Min-John Lee;Ed Matthews; Arrian Mehis; Todd Mitchell; Gajanan Mudaliar; ChandraS. Mugunda; Todd Muirhead; Brian Nelson; Dat Nguyen; Nam Nguyen;Marcos Palacios, Ph.D.; Weimin Pan; Ranjith Purush; Ryan Putman; PaulRad; Ashish Ray; Scott Rosen; Amina Saify; Ananda Sankaran; SpencerSells; Sandhya Senapathi; Michael Smith; Hector Valenzuela; Vicki VanAusdall; Bryant Vo; Chip Webb; and Haihong Zhuo
ART
ART DIRECTOR | Mark MastroianniDESIGNER AND ILLUSTRATOR | Cynthia WebbCOVER DESIGN | Mark MastroianniCONTRIBUTING PHOTOGRAPHER | Joey Pena/Dell Photo StudioCONTRIBUTING ARTIST | Kim Wyatt
ONLINE
WEB PRODUCTION | Brad Klenzendorf
SUBSCRIPTION SERVICES
CIRCULATION | Becky JamesSubscriptions are free to qualified readers who complete the online subscrip-tion form or the subscription reply form on page 89 in this issue. To sign up as anew subscriber, renew an existing subscription, change your address, or cancelyour subscription, submit the online subscription form at www.dell.com/powersolutions_subscribe, return the subscription reply form by surfacemail, or fax the subscription reply form to +1 512.283.0363. For subscriptionservices, please e-mail [email protected].
ABOUT DELL
Dell Inc., headquartered in Round Rock, Texas, near Austin, is the world’s leading direct computer systems company. Dell is one of the fastest growing among all major computer systems companies worldwide, with approximately 47,800 employees around the globe.Dell uses the direct business model to sell its high-performance computer systems,workstations, and storage products to all types of enterprises. For more information, please visit our Web site at www.dell.com.
Dell, the Dell logo, Dell OpenManage, PowerEdge, and PowerVault are trademarks of Dell Inc. in the United States, other countries, or both. Other trademarks and trade names may be used in this publication to refer to either the entities claiming the marks and names or their products.
Dell Power Solutions is published quarterly by the Dell Product Group, Dell Inc. DellPower Solutions, Mailstop 8456, Dell Inc., One Dell Way, Round Rock, TX 78682, U.S.A. This publication is also available online at www.dell.com/powersolutions. No part of this publication may be reprinted or otherwise reproduced without permission from theEditor-in-Chief. Dell does not provide any warranty as to the accuracy of any information provided through Dell Power Solutions. Opinions expressed in this magazine may not be those of Dell. The information in this publication is subject to change without notice. Any reliance by the end user on the information contained herein is at the end user’s risk. Dell will not be liable for information in any way, including but not limited to its accuracy or completeness. Dell does not accept responsibility for the advertising content of themagazine or for any claims, actions, or losses arising therefrom. Goods, services, and/oradvertisements within this publication other than those of Dell are not endorsed by or in any way connected with Dell Inc.
ON THE FRONT COVER | Managing racks of Dell/EMC storage
Network Link Aggregation Practices with the Dell PowerEdge 1855 Blade Server By Bruce Holmes
The networking architecture integrated on the Dell PowerEdge 1855 blade server enables significant flexibility and resource consolida-tion. This article describes the relationship between the optional DellPowerConnect 5316M Gigabit Ethernet switch when configured for link aggregation and the Dell PowerEdge 1855 blade server’s integratednetworking architecture in enterprise networking environments.
Open Source Utilities for Remote Monitoring of HPC ClustersBy Yung-Chin Fang, Randy DeRoeck,Rinku Gupta, and Monica Kashyap
Computing centers continue to scale out in a quest for additional processing power, making remote hardware management critical for cluster hardware uptime. This article provides examples of how opensource utilities such as IPMItool and Ganglia can be utilized and inte-grated into large computing infrastructures to monitor and manage hardware health conditions.
Enhancing High-Performance Computing Clusters with Parallel File Systems By Amina Saify; Garima Kochhar; Jenwei Hsieh, Ph.D.; and Onur Celebioglu
The performance of I/O subsystems within high-performance comput-ing (HPC) clusters has not kept pace with processing and communica-tions capabilities. This article discusses the advantages that parallel file systems can offer HPC clusters and how these specialized filesystems can help enhance overall scalability of HPC clusters.
HPC clusters have evolved from experimental computing architectureto mainstream supercomputing architecture. Today, HPC cluster architecture integrates multiple state-of-the-art, industry-standardcomponents to provide aggregated, cost-effective supercomputingpower. This article examines the influence of component selection on cluster performance and manageability.
need to remotely monitor hardware health conditions
without compromising host CPU cycles, memory space,
and other in-band management components. In response
to this necessity, the computing industry has designed
certain frameworks that primarily have focused on out-of-
band (OOB) management concepts. OOB refers to a man-
agement technique that allows administrators to manage
their systems remotely without consuming host resources.
In the past, most tier-one venders have implemented OOB
management frameworks. Over time, as computing centers
have grown with multiple generations of heterogeneous
hardware, interoperability has become a desirable charac-
teristic to reduce administrative overhead. Thus, remotely
managing a growing and potentially heterogeneous com-
puting facility has become a challenge.
In the past decade, the computer industry has
designed and implemented multiple specifications for
hardware health monitoring and management by using
different management fabrics, processors, and sensors.
These implementations have led to complex interoperability
issues and high management costs, triggering the develop-
ment of an industry-wide, standardized remote hardware
management specification. The most frequently requested
remote management features include OS-independent
remote power management (such as remote power cycling
of a hung node), reduced cost of the remote hardware man-
agement fabric, proactive prevention of failures, and remote
access to motherboard sensor readings. Such requests from
system administrators have prompted the computer industry
to define the Intelligent Platform Management Interface
(IPMI) specification.1
Out-of-band managementOOB management techniques enable administrators to
remotely monitor and manage the health condition of
one or more platforms when an OS is not present. For
BY YUNG-CHIN FANG, RANDY DEROECK, RINKU GUPTA, AND MONICA KASHYAP
Open Source Utilities for Remote Monitoring of HPC Clusters
Computing centers continue to scale out in a quest for additional processing power,
making remote hardware management critical for cluster hardware uptime. This article
provides examples of how open source utilities such as IPMItool and Ganglia can be
utilized and integrated into large computing infrastructures to monitor and manage
hardware health conditions.
1 For more information and technical detail about IPMI, see “Industry Standards for Managing the HPC Cluster Life Cycle” by Yung-Chin Fang and Rizwan Ali in Dell Power Solutions,s June 2004;
and “Managing and Monitoring High-Performance Computing Clusters with IPMI” by Yung-Chin Fang, Garima Kochhar, and Randy DeRoeck in Dell Power Solutions,s October 2004.
example, in the HPC cluster deployment phase, the OOB remote
power management feature can be used to power up a plat-
form—via the OOB network fabric—with no OS. IPMI architects
have designed the baseboard management controller (BMC) as
an embedded and dedicated management processor, including
a set of management buses and a set of sensors and interfaces
to interact with several existing management specifications and
implementations. To help reduce the complexity of network fab-
rics, the IPMI OOB fabric is Ethernet-based and can share the
embedded network interface card (NIC) port on the motherboard.
On some eighth-generation Dell™ PowerEdge™ servers—including
the PowerEdge 1850, PowerEdge 1855, and PowerEdge SC1425
servers—the BMC can be accessed through NIC 1 in this shared
NIC configuration.
In addition to powering on the node to deploy an OS remotely
in the HPC cluster deployment phase or operational phase, admin-
istrators can use the OOB fabric to access the node sensor or system
data record (SDR) remotely. The SDR contains the on-board sensor
readings and preset thresholds. When a threshold is met, the embed-
ded BMC sends a Platform Event Trap to a centralized manage-
ment console. The management console can then e-mail the system
administrator to take action, or it can launch a predetermined appli-
cation to remedy the issue. In the HPC cluster operational phase,
OOB management is also designed to remotely reboot or power cycle
a hung node, remotely access the system event log (SEL), start a
remote console to a node, and—based on the log—automatically
take action to prevent hardware failure. All the preceding features
are implemented via embedded IPMI firmware.
IPMItoolIPMItool is an IPMI 1.5–compliant software utility designed to
manage IPMI-compliant platforms through either a kernel device
driver for execution on the host or through Ethernet by commu-
nicating with the IPMI firmware on a remote platform. IPMItool
is BSD license compliant2 and can be downloaded from www.
sourceforge.net/projects/ipmitool.
The IPMItool command-line interface (CLI) provides the follow-
ing capabilities locally and remotely:
• Obtain field replaceable unit (FRU) information3
• Identify (get) or set the local area network (LAN)
configuration • Obtain sensor readings • Access the SEL • Issue IPMI-based power management commands
The Figure 1 example
shows IPMItool sensor read-
ing output from one node.
IPMItool output is thorough
but not easy to analyze
when data centers have a
large number of nodes. Addi-
tionally, most data centers
have limited administrative
resources. In this case, an
automated sensor monitoring
mechanism is needed. This
article examines a mechanism that is designed to present periodic
and large amounts of sensor data in a user-friendly way
GangliaGanglia is an HPC system monitoring tool that grew out of two
clustering research projects at the University of California, Berkeley:
The Millennium Project and its predecessor, the Berkeley Network
of Workstations (NOW) project. The Millennium Project team has
also worked closely with the National Partnership for Advanced
Computational Infrastructure (NPACI) Rocks group at the San Diego
Supercomputer Center (SDSC). In fact, Ganglia was introduced at
the SuperComputing 2000 conference at the NPACI booth.
NPACI Rocks is one of the most popular open source cluster
computing software stacks, and it currently includes Ganglia in its dis-
tribution. Ganglia can be downloaded from ganglia.sourceforge.net.4
It monitors 25 metrics on every node in a cluster, including the number
of CPUs; CPU speed; and percentages for each CPU state—user,
system, nice, and idle. Ganglia operates on the internal cluster fabric,
also known as the in-band management fabric, which utilizes the
cluster’s communication resources.
Central to the Ganglia monitor is the Ganglia Monitoring
Daemon (gmond), which runs on every node in a cluster. All moni-
tored metrics are sent to a cluster-wide multicast channel to which
all nodes subscribe. Monitored data is exported in XML format,
which allows various XML engines to be used or built to display
the exported data in graphical form. Ganglia uses a fully distributed,
push-based approach to monitoring—that is, any node in the cluster
can be queried for cluster state information and such information
is available on every node at all times.
Enhancing Ganglia with gmetric By default, Ganglia displays a limited number of metrics. Although
these metrics are meaningful for monitoring OS activities, Ganglia
2For more information about BSD licenses, see www.opensource.org/licenses/bsd-license.php.
3Most tier-one venders like Dell have FRU information coded on firmware for major components such as the motherboard.
4For more information about Ganglia, visit ganglia.sourceforge.net/docs.
OOB management techniques
enable administrators to
remotely monitor and manage
the health condition of one
or more platforms when
an OS is not present.
Check the Dell Power Solutions Web site for sour late-breaking exclusives, how-to’s, casestudies, and tips you won’t find anywhereelse. Want to search for specific articlecontent? Visit our Related Categories index online at www.dell.com/powersolutions.