H3C Technologies H3C Intelligent Management Center - Manual (Page 45)

31
Having baseline data enables you to measure the effect that IMC has had on supporting the
realization of the organization's goals and aligning IT with business objectives.
2. For What Timeframe?
• Is the data for alarming purposes? Is the data needed in near real time? If so, what are the service
level commitments for performance, outage duration, or service availability? How long can the
organization wait before being notified of a problem or outage?
These questions identify some of the monitoring requirements for fault or event management.
Fault management identifies current failures or problems that need immediate attention. The
collection and processing of data and event notification for fault monitoring need to align with
the service level commitments that stakeholders make. For example, a service level commitment
of 99.999% network reachability translates into about five minutes of network outage per year.
The polling interval used for monitoring for reachability must be significantly less than five
minutes in order to support the achievement of this service level commitment.
• Is historical reporting required? If so, in what areas is historical reporting needed? On resource
performance? On faults or events in the network? On device configurations and changes to
configurations? On network assets and changes to network assets? On security events? On access
to IMC and changes to IMC?
These questions identify in what areas historical reporting is needed. Historical reporting can
encompass everything from network and performance management to asset, configuration,
change, incident, problem, and service level management. In fact, most groups require some
form of historical reporting and the key is to provide stakeholders with reports targeted to meet
their needs.
• What level of data granularity is needed? How much summarization is needed? For how long does
data need to be retained?
The first question can be translated to how frequently data must be polled from the devices, a
key requirement as it defines how you configure polling in IMC. How many data samples does
the stakeholder need to reliably measure for their key performance indicators? Once you have
this information, you can evaluate the load that polling places on the:
− Resources being monitored.
− Network resource and bandwidth consumption that monitoring introduces.
− The processing and storage load that places on IMC. Ultimately, this involves finding a
balance between stakeholder needs and network and IMC resource consumption.
Data retention and summarization requirements vary for different stakeholders. The on-call
network operations team usually requires data retention with less summarization, typically one
day to one week of data retention. Less summarization or more granularity is needed for
network operations teams as they are most closely focused on pinpointing when, how, and
why problems occurred. The data retention and summarization requirements of performance
analysts whose time horizon may be up to one month or more fall in a mid-range. Capacity
planner time horizons are much longer and may span five years and with much greater data
summarization.
Data retention is another matter for asset, problem, incident, change and configuration
management as both detail and summary reports are required. In such cases, data export
options may be considered.
3. In What Format?
• If the data is needed for alarming purposes, what method is used? Email or SMS text notifications?
Console display of events and alarms? Integration with other management systems? Integration
with Help Desk ticketing systems?