Compaq ProLiant 1850R, 6400R - ProLiant - 1 GB RAM User Manual
HARVARD RESEARCH GROUP
COMPAQ’S CLUSTER MANAGEMENT INITIATIVE
Harvard Research Group has determined through extensive primary and secondary
research that today’s business managers want Enterprise Management Systems that can
help them gain control over the business process and reduce complexity. IT managers are
increasingly looking for ways to effectively and efficiently implement new technologies
while at the same time achieve a seamless and cost effective way to manage their systems
and software. They want to "virtualize" their system environment and manage it as if it
were a single logical system. At the same time, the rapid growth in the number of
business-critical applications running on Intel-based systems coupled with the need for
high availability and continuous data access is driving the demand for clustered servers.
While clusters help solve problems with availability and scalability, they also bring
increased complexity, immature management tools, and added cost. This situation is most
evident in the Intel server market where the use of Microsoft Cluster Server (MSCS) is
accelerating faster than the capability of the tools that manage it.
Harvard Research Group has spoken with more than 700 users of highly available
systems. Our findings and research show that users of high availability systems typically
have mixed operating system and hardware environments, and believe that current cluster
system management tools are too complex and costly. They view the current set of
cluster administration tools as point products that are difficult to integrate, proprietary in
nature, and require too much time and money to properly implement. Many of them have
expressed the desire to reduce the complexity of cluster configuration, administration,
and problem determination / resolution. At the same time, they have expressed a desire to
minimize the number of interfaces and tools they need to manage their system and
clustered environments.
Harvard Research Group believes that all clustered solutions, still lack adequate and
extensible cluster monitor / management tools. While there are management tools that
partially address this issue for a particular operating environment (HP/UX, Solaris, NT),
they tend to be limited in their functionality and unable to address the larger issue of
managing multiple and dispersed cluster environments. This is contributing to the high
cost of ownership of server hardware and causing potential profits to be consumed by
increased administrative staff and lower user productivity. Studies have shown that
deploying manageable systems in concert with well managed IT processes can create
actual value for a company. It stands to reason, then, that tightly integrating cluster
management tools with the non-clustered systems management interface should create
more value.
It appears that Compaq, in concert with its acquisitions of Tandem Computer and Digital
Equipment Corporation, is building itself into an enterprise systems provider with some
key strides in changing the direction of enterprise systems management. In doing so,
Compaq has defined some interesting goals and is delivering on these goals.
Compaq’s E2000 Platform Architecture and Cluster Management Vision
HARVARD RESEARCH GROUP
Copyright 1999 Harvard Research Group. Reproduction is forbidden unless authorized.
For additional copies call (978) 263-3399
Box 297 HARVARD, MA 01451 USA
http:/www.hrgresearch.com
Compaq's E2000 Platform Architecture is based around industry standard servers,
networking, and storage systems that are connected via high speed SAN interconnect to
create a scalable and available clustered server environment. Some of the significant
components of the E2000 Platform Architecture include:
2-, 4-, & 8-processor Compaq ProLiant Servers
Intel 32-bit Pentium Pro, Pentium II, Xeon, and 64-bit Merced servers
Enterprise Management Utilities for large scale application processing
Fibre Channel-based storage products and backup libraries
Gigabit Ethernet, ATM and other high speed networking technologies
High availability, scalable System Area Network SAN interconnects
The E2000 Platform Architecture would integrate these components with industrystandard operating systems such as Microsoft Windows NT, Novell NetWare, and SCO
UnixWare and with key enterprise applications from Microsoft, Oracle, SAP and other
ISV’s. Many of the key Compaq components of the E2000 Platform Architecture are
already being delivered to customers (such as high-availability SMP servers and
ServerNet SAN interconnect), while others like integrated and easy-to-use management
and administrations tools are just beginning to take shape as part of Compaq’s Cluster
Management Vision. By tightly integrating leading, standards based, enterprise
management tools with their E2000 Platform Architecture and Cluster Management
products, Compaq will be able to provide users an easy-to-use, end-to-end solution that
supports easy administration and troubleshooting of all the components of an enterprise’s
IT structure.
Compaq’s Cluster Management Vision
Compaq has outlined a cluster management vision and road map that will allow them to
deliver intuitive, easy-to-use cluster management tools that simplify the tasks of
planning, deploying and operating clusters. The Tables below highlight Compaq’s cluster
management concepts, goals and how they will bring high-end cluster system
management features to the volume cluster market and help develop industry standards.
Ease of Use
Common data repository for all enterprise devices
Problem definitions and remedies for rapid resolution
Extensive help text for ease of understanding
Web based management tools for flexibility and efficiency
Intuitive and consistent User Interface for immediate
understanding
Richness of Function
Common Data Repository
Real Time Cluster Monitor
Cluster Administration and Configuration Management
Extensible Architecture
Multi Node Cluster Support
Common Cluster Monitor and Administration Utilities
HARVARD RESEARCH GROUP
Box 297 HARVARD, MA 01451 USA
2
http:/www.hrgresearch.com
Application Capacity Modeling
Dynamic Load Balancing
Self Defining Clusters
Application Policy Management
Extensibility
NT – MSCS Proliant Cluster as a base
Industry Standard Server and Operating System Support
Digital and Tandem Legacy System Support
The management and administration redundancies caused by having to support six
different operating system environments and two processor architectures has compelled
Compaq to develop a strategy and plan for implementing a consistent cluster
management environment. Compaq intends to leverage Tandem and Digital technologies
to deliver enterprise cluster management functionality to the NT volume cluster market
and provide industry standards for cluster management. While initially they are
implementing their cluster management tools for NT and Microsoft Cluster Server
(MSCS), they fully recognize the need to include all their supported operating
environments. Over time Compaq is expected to enhance its cluster management tools to
include it’s Operating System partners Novell NetWare, SCO UnixWare, as well as
Compaq legacy systems Tru64Unix, Open VMS, and Non-Stop (NS) systems.
The first tangible evidence of this strategy is two new NT MSCS management offerings.
These are Cluster Monitor and Intelligent Cluster Administrator. They leverage the
power of anytime and anywhere Web technology and operate within Compaq’s Web
based Enterprise Management (WBEM) framework.
Compaq Insight Manager XE - Cluster Monitor
The Cluster Monitor is a management subsystem that allows administrators to view all of
their clusters from a single browser. As a subsystem of Insight Manager XE, it is fully
integrated into Compaq's Insight Manager suite of offerings, thus providing a one-stop
administrative facility helping to minimize the number of applications and interfaces
required to manage the systems environment. It supports the ability to configure
monitoring points and specific operational performance thresholds, which will alert the
administrator when critical cluster operational thresholds have been met or exceeded.
Administrators can select cluster configurations based on their level of responsibility,
location, or cluster type, as well as perform administrative tasks either locally or remotely
from the cluster. Cluster events can be prioritized allowing the administrator to select an
event and drill down to get a view of the specific cluster node where the problem exists.
Some of the more significant features and benefits of Compaq’s Cluster Monitor include:
Web based interface for ease of utilization with anytime anywhere access
Customizable to meet administrator workload or job responsibility
Selectable monitor points for cluster monitor customization
Security levels for administration control and access
Problem Window and Topology Tree event selection for micro or macro management
Event Problem Description and Potential Remedy displayed for efficiency
HARVARD RESEARCH GROUP
Box 297 HARVARD, MA 01451 USA
3
http:/www.hrgresearch.com
Drill down to problem device for further analysis and repair
Extensible for third party management system provider interfaces
The Cluster Monitor, to be delivered as a subsystem of Compaq’s Insight Manager XE,
will be included with all Compaq servers. A major benefit to system administrators is its
single point of control for desktops, servers and clusters. The browser-based interface
provides administrative flexibility via anytime anywhere access making cluster
management less complex and time consuming. It is both simple and easy to view, with
extensive problem definition/remedy information displayed whenever an event or device
is selected. The intuitive color-coded identification of cluster "hot spots" supports rapid
problem determination and resolution that can make the administrator more efficient,
decrease cluster downtime, and minimize "hung" systems.
Compaq Insight Manager XE – Cluster Monitor employs Compaq management agents
for basic information on system health and the health of the MSCS cluster. Compaq has
structured the Cluster Monitor to utilize the Insight Manager XE Event and Data
Repository for data storage. As a result, the cluster monitor will automatically perform an
aggregation and presentation of that data to the administrator, who can then review it and
take whatever actions are appropriate.
The administrator can easily access the Cluster Monitor through a query from the Insight
Manager XE introduction screen. The administrator can define multiple queries to view
different cluster environments for flexibility and ease of use providing another level of
efficiency. These queries can be structured to meet a variety of administrative needs, such
as, all clusters, clusters performing a specific set of applications, clusters on the tenth
floor, or clusters in the state of New Hampshire. The first thing the administrator sees
after navigating through the introduction screen is a display that presents critical cluster
events and a tree structure reflecting those clusters as defined by a specific query.
Anytime an event occurs that is considered destabilizing, an alert notice will be broadcast
to the cluster monitor screen. Alert notices that require immediate attention by the
administrator are likewise added to the Cluster Problem Window and highlighted in the
Topology Tree with a color-coded severity notation.
HARVARD RESEARCH GROUP
Box 297 HARVARD, MA 01451 USA
4
http:/www.hrgresearch.com
Loading...