Systems And Network Management

Fast access to knowledge about each PC in
the enterprise and tools to manage tasks are critical factors in managing these valuable
resources cost-effectively.

Systems and network management tools have
evolved along different paths that are covering to provide a unified solution to current
management needs. Systems management developed from a mainframe perspective; network
management responded to the need to manage telecommunication systems and, eventually,
datacom systems such as WANs and LANs. And understanding of both areas is necessary for
today’s combined systems and network management requirements and solutions.

Enterprise management tools
Enterprise management is the collection of activities that involve configuring,
controlling, monitoring, analyzing, diagnosing, repairing, operating, supporting and
securing all components of a networked computing environment. It includes many of the same
aspects of traditional mainframe computing such as storage or security management and some
additional elements associated with distributed computing.

Enterprise management tools can be
proactive or reactive, accessible through the web or a graphical user interface (GUI) and
used by internal information systems departments or third-party management services.
Enterprise management tools include those used for monitoring, controlling and planning.
Monitoring provides status information on all elements; controlling facilitates response
to information obtained from the monitoring process. Planning uses the information
obtained in monitoring and the decisions implemented in controlling to predict future
needs and prepare for change requirements.

The following list reflects elements of
traditional mainframe and distributed computing enterprise management tools:
ACCOUNTING: Computing and network resources represent a
significant investment, encouraging companies to track assets and inventory and monitor
resource usage. Efficient accounting enables administrators to optimize purchase decisions
and to charge other organizations appropriately for network use. Automated data collection
tools allow administrators to focus on keeping networks operating smoothly and set aside
time for proactive planning.
AUTOMATED OPERATIONS TOOLS: These
tools automatically perform activities such as writing a file to a tape system. They are
extremely sophisticated and include intelligence that supports actions such as rerunning a
task if it is interrupted.
BATCH PROCESSORS: Batch processors collect a series of
transactions and process them as a batch. A company may gather payroll transactions during
the day and process them at night when a computer has a lighter workload. Batch processors
help maximize systems resource use and contribute to a more efficiency reporting process.
CAPACITY PLANNING TOOLS: Capacity
planning determines and projects how much computer power a company needs. These tools
examine the amount of processor, memory and disk storage an application will require
before it is deployed; help determine whether upgrades are needed to run the application
and facilitate decisions about capacity expansion. Planning involves forecasting loads,
adjusting for growth and incorporating new technologies as necessary. Trending data and
data modeling are key components of successful planning. Network and system captivity
planning is critical because changes in business requirements must be supported
immediately by the corporate network. Accurate planning enables administrators to respond
to new demands effectively without burdening existing resources.
CONFIGURATION MANAGEMENT: Managing
a network configuration involves collecting configuration information about all enterprise
resources, including systems, servers, network elements and devices. This information
typically is stored in a database, where systems and devices are associated with physical
and logical network segments. Configuration changes are logged into this database to
facilitate troubleshooting, detect unauthorized changes and assist in inventory
management. Configuration management tools ensure that configuration data is always
available and accessible.
DISASTER RECOVERY: Disaster recovery comprises the
procedures, policies and products that allow for the timely resumption of the
computer-based elements of an enterprise’s business processes following a significant
large-scale interruption in service. The fundamental assumption behind a disaster recovery
plan is that the site where the computers reside either is not usable or not accessible.
Disaster recovery is a systems management discipline allied with but separate from
security and data availability. Although the consequences of a security breach certainly
could be disastrous, security should be treated as a distinct systems management topic of
its own. Moreover, although backup and duplicate hardware can improve data availability,
the existence of a backup tape is only one part of successful recovery from a disaster.
DISK BACKUP SYSTEMS: These tools copy information from
disks to tapes and other backup systems. They identify where files are located, may
compress information so that it is stored efficiently and may include mechanisms to
restore corrupted or deleted files.
EVENT MANAGEMENT: Information supplied by event
management systems can include events such as the addition of a new system or router, an
early warning of crossing a performance threshold or notice of a fault such as a device
failure. Some problems can be solved with automatic, built-in procedures while others are
routed to designated staff. Event management is an important reactive tool for network
administrators and it also provides information useful for future planning. Some network
management products integrate both event management and fault management.
FAULT MANAGEMENT: Fault management involves recognizing
alert conditions, generating alerts, and forwarding alerts to the appropriate management
resource. This area may include trouble ticketing systems, help desk tools, historical
problem data and measurements of problem solving effectiveness, such as mean time to
repair. Efficient fault management ensures rapid problem response and assist
administrators in assessing why faults occur and preventing future instances.
JOB SCHEDULERS: These tools control traffic and
determine when specific jobs can use a resource such as an operating system. Job
schedulers establish procedures to ensure that high-priority transactions are processed
quickly. They can move a mission-critical application, such as an airline reservation or
funds transfer, ahead of a less critical application.
NETWORKING MONITORING: Network monitoring tracks
network statistics at the port, line and device level. Tools can notify the management
applications whether workstations and servers are available and supply port and device
status, configuration views, trends analysis for problem prevention, alarms on events,
bandwidth utilization, error detection, diagnostics and power supply monitoring.
Comprehensive network monitoring helps administrators maintain an overall view of network
health.
PERFORMANCE MONITORING TOOLS: Performance monitoring
examines how well a system is operating and determines the location of any bottlenecks.
Such tools may direct a company to install more internal memory to ensure, for example,
that all transactions will be completed in less than five seconds. Enterprise performance
management collects performance data on all areas of a network and triggers alarm when
thresholds are exceeded. Responses to alarms may include load balancing by reassigning
applications to appropriate processors and tracking applications to ensure availability.
Historical logging and trend analysis may be used for performance planning. Performance is
a key indicator of network and system efficiency and often correlates with customer
satisfaction.
PRINT MANAGEMENT: Print management tools route
documents to appropriate print resources, manage print queues and monitor printer
functions. Documents can be routed to an alternate printer if a problem arises. In many
situations, print servers facilitate printer sharing and maximize print throughput. Print
servers can be dedicated hardware connected directly to the network, software resident on
a network PC, or a card that plugs into the printer.
SECURITY MANAGEMENT: These tools provide remote access
monitoring notification and management of systems that authenticate users, authorize their
requests to use system resources and audit the use of these assets. Most operating systems
include basic features such as password security, but more sophisticated security is
available through add-on modules that provide flexible assignment of application to
individual users and selected groups. Data collected by security systems includes access
records that provide detailed information on log-in attempt locations and times and on
modem activity for remote access. Automatic logging and responses to security violations
are also essential features of a security solution.
STORAGE MANAGEMENT: Storage
management enables companies to use available disk, tape and other storage systems
efficiently. These tools enable administrators to view, define, sort, filter and analyze
current storage information. They feature highly available, redundant storage or
intelligent caching as well as uniform naming systems so data is stored in a consistent
manner that facilitates retrieval. Third-party solutions are now available to manage
network data backup and recovery and simplify the administration of enterprise-wide
storage management. These systems provide automated backup of network-based corporate data
and a range of solutions for data location and retrieval. Using a variety of storage
technologies, including traditional disk, magnetic tape and optical disk, these systems
offer capabilities such as tracking data usage and maintaining most used data on disk and
least used data on tape.

Reducing total cost of ownership
A key aspect of developing tools for distributed environments is meshing
mainframe class systems management features with the open, standards-based environment of
desktop systems and servers. Such tools are needed to improve user and administrator
productivity as well as reduce the total cost of ownership of distributed systems. Gartner
estimates that an unmanaged PC running Windows 95 has an average annual cost of ownership
of $9,784. With today’s corporate networks commonly encompassing thousands of such
systems, even small reductions in per system cost of onwership can be significant.

Reduced cost of ownership has been a major
marketing emphasis for vendors of network computers (NCs), who essentially remove from the
PC the elements that require management. Data, software backups, upgrades and maintenance
are received from a network server. NCs are designed to be almost maintenance free and
lack the requirements for disk storage or high-end computing power, making them an
economical alternative to PCs. Some major players in the NC arena include IBM, Oracle
(which offers an NC product through Network Computing), and Sun.

Although the NC may provide a solution for
reducing cost of ownership in the future, the number of network computers that actually
have replaced PCs is not great enough to have a significant impact on the cost of system
onwership in large enterprises. NCs may offer a better replacement for terminals, which
are already relatively low cost.

One effect of the marketing efforts of NC
vendors has been an increased awareness of PC cost of ownership. The response from PC
hardware and software vendors has been a greater emphasis on reducing cost of ownership
and offering zero administration desktops. In 1997, Microsoft and Intel announced a
diskless PC standard, called the NetPC, that is in between a thin client and a PC to help
address concerns about high PC ownership costs. Compaq, Dell and HP collaborated in
developing the NetPC specification. Companies will benefit from these initiatives because
they serve to lower PC costs over time.

Microsoft’s Zero Administration for Windows
(ZAW) initiative includes features designed to reduce the cost of managing desktop PCs,
including automatic installation of operating system updates when the computer is turned
on, automatic installation of applications as they are invoked by the user; storage of
user files and profile information on a server rather than on the local hard disk and the
ability for a central administrator to specify system configurations that cannot be
modified by the user. Microsoft’s Zero Administration Kit (ZAK) was the only component of
the company’s ZAW initiative available in late 1997 and was developed to complement
Microsoft’s System Management Server (SMS) software. ZAK provides for the management of
users through policies. SMS offers automated software and hardware inventory, software
distribution and remote diagnostics.

During 1997, Intel continued to build on
its Wired for Management initiative, which aims to bring manageability features to
conventional PCs in a consistent way. HP announced plans in October 1997 to integrate
features based on Intel’ Wired for management specification into its line of Vectra PCs
and commercial workstations. The specification will give users standard instrumentation,
remote boot ROM and PC wake-up features.

HP works with other systems manufacturers
to reduce the cost of networked computing through a program known as HP OpenView-Ready to
complement HP’s management platform, HP OpenView. Jointly developed by HP and Dell (also
the first program participant), the HP OpenView Ready program includes the HP OpenView
Ready Network Node Manager at no additional cost when customers purchase selected Dell
PowerEdge servers running Microsoft’s Windows NT or Novell’s IntraNetWare. The OpenView
Ready Network Node Manager allows network administrators to monitor the health of their
network servers and management parameters of their servers, desktops workstations,
notebook computers and network devices. This information is integrated with HP OpenView.

Companies have a stake in making computers
more manageable in interconnected enterprise environments, using management standards such
as the simple network management protocol (SNMP), windows management instrumentation (WMI)
and web-based enterprise management (WBEM). With less money required for management,
customers’ finances are available for investing in new technology and network growth
instead of maintaining existing technology.

Network management
Today’s corporate networks commonly may encompass thousands of components,
including mainframe systems, servers, desktop and notebook computers, wireless devices
such as smart phones, printers and networking devices such as routers and switches.
End-user systems and network devices are connected via LANs that carry information at high
speeds over short distances. LANs in turn are connected to high-speed backbone networks
that interconnect via WANs.

With the advent of client server computing
in the mid-eighties the issue of LAN management increased steadily in importance. From an
environment with few tools and where LAN users were responsible for backing up their own
data to floppy disks, LAN management has become extremely sophisticated. Till recently,
however, LAN and distributed systems management tools focused on hardware monitoring
performance, providing status information (such as whether a system of a specific port was
available) and implementing device control. LAN management tools now play a significant
role in applications management. These tools are capable of performing a wide range of
tasks, such as determining whether a particular instance of database software is available
at a particular server.

Within this environment, the emphasis is on
managing applications end-to-end, which typically encompasses not only LAN traffic but WAN
traffic as well. Issues such as software distribution, configuration management and
security control are just as important as traditional operational systems management. In
addition, the inherent size and complexity of the corporate network demand expanded
capabilities. Tasks such as capacity planning can be far more complex for a heterogeneous,
distributed enterprise network because of the potentially large number of factors
associated with applications, users and network media.

With recent advances in standards such as
additions to the remote network monitoring specification (RMON2) network management tools
are capable of addressing every element of the enterprise network and they have become
robust enough to rival those found in legacy mainframe environments. The tools used to
manage both LANs and WANs are based on SNMP and offer similar capabilities.

These tools have helped administrators
control the distributed computing architecture of enterprise networks encompassing
mainframes and distributed systems in a mixed LAN and WAN environment. However, a new
networking paradigm has exploded onto the networking scene: internet-based web technology
and corporate intranets (An intranet is an enterprise network based on internet
technology). Internet technology includes a standards-based network that makes heavy use
of web site and browser technology to provide easy access to information. The internet has
contributed to the convergence of systems, network and application management and it is
generating a need for new management tools designed to take advantage of web technologies.
Intranets also have intensified existing networking trends, such as increasing the number
of applications on the network, the need for end-user access to these applications and the
need for administrator access to management data about each application. Extranets
(portions of an intranet accessed by external partners, customers or suppliers) also pose
security, management and planning challenges.

Self-healing networks
Tools are evolving to meet management needs and are moving toward high levels of
automation, with an eventual goal of enabling self-healing networks. Components of this
solution will include higher levels of local intelligence throughout the network in the
form of intelligent agents and policy-based network management applications that work
proactively to prevent significant network problems. The combination of advanced network
monitoring and distributed management intelligence will allow network management
applications to take immediate corrective action without network administrator
involvement.

Large global networks have many thousands
of entry points and to probe each one would yield an overwhelming amount of information.
Tools that help ensure service quality for groups of users mean fewer experts are required
to maintain complex internetworking configurations.

Traditional offerings frequently present
and track information about service levels. However, many products report performance
statistics only or still require a great deal of operator intervention. Others may show an
alarm about a violation, but they may not show customers what went wrong. Although routers
and switches have become more reliable, recovery processes have become more automated, and
new products offer improvements, widespread implementation of technology that promises
end-to-end network reliability has not yet occurred.

Network management tools from 3Com and
Cisco announced in 1997 illustrate the direction in which the industry is heading. 3Com
also plans to guarantee service levels for specific users. The company announced plans in
1997 to integrate its Transcend management tools with service-level monitoring software
from InfoVista to enforce service policies for traffic moving through its internetworking
equipment.

Cisco’s Netsys Service level Management
(NSM) automated software offers self-healing capabilities for networks with Cisco
equipment. Cisco acquired Netsys Technologies in November 1996 and modified Netsys router
modeling software to add features that will let network managers guarantee service
quality. NSM software works with Cisco’s Catalyst 5000 Lan switches and with routers from
Cisco and Bay Networks. (Cisco also plans to support StrataCom’s WAN switches).

NSM is designed to help customers predict
where problems will happen and react to failures. NSM verifies whether enough bandwidth is
available on network circuits before multimedia traffic is added and analyzes circuit
integrity to help minimize connectivity problems. NSM helps monitor network performance,
diagnoses problems and suggests possible solutions. It also helps define service policies
and track end-to-end performance network-wide.

Issues in systems and network
management
Just as the client server paradigm forced a radical change in management systems,
internet and web technologies are changing the management rules again. At the same time,
more proactive approaches to management are being pursued. For example, the network
maintains a given performance level by predicting and quickly reacting to anticipated
network events accurately. This view is reflected in many vendors’ strategies as an
eventual goal of ‘self-healing’ networks. Some products to support this goal already are
available in the form of policy-based management applications and embedded intelligent
agents. However, developers of management tools are just beginning to implement this type
of automation.

Using web-based management tools can yield
operational advantages and potential cost savings. Web-based management products provide a
universal client that enables relatively simple, geographically dispersed, multi-user
access to management functions and information. These solutions range from simple, read
only access of network monitoring data to interactive tools that support advanced
management.

Distributed access to advanced management
capabilities offers the freedom to solve network problems from home or while traveling.
Technical end-users benefit from access to network status information as well, eliminating
a call to a busy support desk or planning work to avoid peak use times when the network is
congested.

One significant aspect of web-based
management is platform independence. Users with web access can obtain information with
equal ease and without special set-up. Development costs and time-to-market are reduced
substantially because only a single server-based application accessible from all platforms
is required. Platform independence also means web-based management systems can monitor all
elements of the enterprise network, including LANs, WANs and telecom systems. Even
peripherals such as printers now offer browser-based access to management information,
allowing users to manage their time and productivity better by making use of these
resources.

Load balancing and the web
Load balancing and performance management of web traffic have become bigger
issues as the number of web users has grown steadily. Traditional methods using a Domain
Naming System (DNS) server allow for situations where some servers may be overloaded and
others idle. Incoming traffic may be sent to a web server without regard to that server’s
existing load.

Offerings such as Cisco’s Local Director,
HydraWeb’s HydraWeb Load Manager, IBM’s Network Dispatcher and RND Networks’ web server
Director Pro help ISPs, corporate intranet managers and large web site operators increase
reliability and performance of a web site. These offerings helps regulate traffic between
the internet and a group of servers. Proxy server standards are being developed that may
help optimize performance as well.

Cisco’s Local Director is a high-end
hardware and software offering that has several options for distributing traffic. Servers
can be brought online gradually, and load is based on server response time.

Hydraweb’s software and software/hardware
products offer load balancing, monitoring, management and fault-tolerant features. The
Load Manager product can send out alerts via email, fax or pager. Servers also can be
managed remotely using token-based authentication.

In September 1997, IBM and its Tivoli
Systems subsidiary introduced Java-based tools for managing enterprise hardware, networks
and web-based business systems. As part of this announcement, IBM’s Internet business
launched interactive network Dispatcher 1.2, a low-end, Java-enabled, load balancing tool
for web servers.

RND Networks Web Server Director uses a
combination of load balancing techniques (similar to Cisco’s Local Director), and provides
some security, alert and redundancy features; however, in its 1997 version, it does not
support 100Mbps Ethernet.

Intermediary or proxy servers also are
being used to increase performance. Proxy servers store data that individuals request from
the internet. Requests for this same data later then can be served locally rather than
going back over a congested network. With the existing Internet Caching Protocol (ICP),
proxy servers communicate with each other to see whether another server already has the
requested data before searching the internet. Using this protocol may be helpful but
servers may be storing redundant data. Also, a large group of proxy servers generates more
queries between servers, which creates more overheads. A new proposed standard, the Cache
Array Routing Protocol (CARP) by Microsoft and Netscape, uses scripts to maximize
efficient query routing. Loads can be balanced better among servers through the use of
simplified routing and configurable load factors.

Applications management
As systems management tools evolve to manage in distributed environments, the
emphasis on traditional areas such as performance management has expanded to include
application performance. The discipline of applications management is based on the concept
that the bottomline for business management is not whether the systems and networks are
running at high-performance levels, but whether the business applications themselves are
available with the necessary performance to meet relevant business needs. Ideally,
applications management should include administration, availability, life-cycle
management, performance, process automation and recovery from unscheduled events.

Traditional systems management metrics such
as system latency or the number of network segment packets dropped can provide a
misleading perception of application performance efficiency. These indicators measures
attributes of the system that delivers services to clients, but they do not indicate the
Quality of Service (QoS) actually realized by the end-user. To determine QoS, metrics must
focus on meeting user expectations of availability, performance, accuracy and
affordability.

Determining whether applications meet
relevant business needs first requires that those needs be quantified by end-users and
systems administrators. For example, a hospital’s accounting department may find a given
response time acceptable, but doctors using an online diagnostics system might find the
same response time unacceptable. Performance numbers no longer can be treated simply as
numbers but must be examined with regard to what those numbers represent in terms of
appropriate service to the user.

Once appropriate application requirements
are determined, administrators must have suitable tools to ensure that those requirements
are met. Defining and addressing these needs, categorized as quality of service
management, can be addressed by service level agreement (SLA) tools. SLA software helps
information systems departments ensure that application performance and availability mesh
with business needs.

One popular SLA product is InfoVista from
InfoVista. InfoVista addresses both the creation and implementation of service level
agreements. Using InfoVsita, administrators prepare predefined or customized QoS reports
to ensure clients are receiving needed services. The reports are also useful for the
information systems departments themselves to ensure they have the resources they need to
provide the services and service levels required by end-users.

Some application-specific management tools
are also available. Tivoli Systems’ Application Management for Notes and Domino monitors
the health of the Notes/Domino application across a network with up to hundreds of Notes
servers and thousands of Notes clients. The application includes systems administration
capabilities for Unix and Windows NT machines and allows security policies to be set based
on policy regions and roles for scalability.

BMC Software has taken a different approach
with its Patrol Series. Rather than provide extensive management for a single application,
the Patrol Series uses ‘knowledge modules’ that focus on the performance and availability
of applications running under a variety of operation systems, database management systems,
key business applications and middleware products.

A key technology used in application
management is the RMON2 standard. RMON2 provides application-specific monitoring rather
than the device port of network segment monitoring provided by RMON. RMON2 provides
information from an application perspective, such as which users are using which
applications and what percentage of available bandwidth each application uses. With
information from RMON2 administrators can manage network resources to ensure that critical
applications are performing at necessary performance levels and that people have access to
the applications they need.

Excerpted with
permission from
Technology Forecast 1998
© Price Waterhouse Associates.

Leave a Reply

Your email address will not be published. Required fields are marked *