Professional Documents
Culture Documents
Lecture#3: Capacity Management: Sana'a Community College CNET Dept. 2 Year Module: Tactical Processes of ITSM
Lecture#3: Capacity Management: Sana'a Community College CNET Dept. 2 Year Module: Tactical Processes of ITSM
Lecture#3: Capacity Management: Sana'a Community College CNET Dept. 2 Year Module: Tactical Processes of ITSM
Headlines
Introduction
Definitions
Basic Concepts
Objectives
Benefits
Process
Relationship with the other Processes and Functions
Activities
Process control
Costs
Problems
Introduction
Good capacity management eliminates panic buying at the last minute, or buying the biggest
box possible and crossing your fingers. Both of these situations are costly.
Definitions
Capacity Management aims to provide the required capacity for data processing and
storage, at the right time and in a cost effective way. It is a balancing act.
Basic Concepts
Important concepts in Capacity Management include:
Performance Management: measuring, monitoring and tuning the performance of
IT infrastructure components.
Application Sizing: determining the hardware or network capacity to support new
or modified applications and the predicted workload.
Modeling: using analytical or simulation models to determine the capacity
requirements of applications and determine the best capacity solutions. Modeling
allows various scenarios to be analyzed and the "what-if" questions addressed.
Capacity Planning: developing a Capacity Plan, analyzing the current situation
(preferably using scenarios) and predicting the future use of the IT infrastructure and
resources needed to meet the expected demand for IT services.
Objectives
Capacity Management aims to consistently provide the required IT resources at the right
time (when they are needed), and at the right cost, aligned with the current and future
requirements of the customer.
Thus, Capacity Management needs to understand both the expected business
developments affecting the customer, as well as anticipating technical developments.
The Capacity Management process has an important role in determining returns on
investment and cost justifications.
Benefits
The advantages of Capacity Management are:
Reduced risks associated with existing services as the resources are effectively
managed, and the performance of the equipment is monitored continuously.
Reduced risks associated with new services as Application Sizing means that the
impact of new applications on existing systems is known. The same applies with
respect to modified services.
Reduced costs, as investments are made at the appropriate time, neither too early nor
too late, which means that the purchasing process does not have to deal with last-
minute purchases or over-purchases of capacity well in advance of when they are
needed.
Reduced business disruption through close involvement with Change Management
when determining the impact on the capacity and preventing urgent changes
resulting from incorrect capacity estimates.
More reliable forecasts the longer Capacity Management is used, which means that
customer requests can be responded to more quickly.
Higher efficiency as demand and supply are balanced at an early stage.
Managed, or even reduced, capacity-related expenses as the capacity is used more
efficiently.
Process
Capacity Management has three sub processes, or levels of analysis where capacity can
be considered:
1. Business Capacity Management – the objective of this sub process is to
understand the future user needs. This can be done by obtaining information from the
customer, e.g. from strategic plans or by undertaking trend analysis. This sub-process
is primarily proactive. It has strong links with Service Level Management in terms of
the definition and negotiation of service agreements.
Capacity Management supports Problem Management in both its reactive and proactive
roles. Capacity Management tools, information, knowledge and expertise can be used to
support Problem Management at various stages.
Capacity Management can be part of the CAB. Capacity Management can provide
information about the need for capacity and the potential impact of a change on the
provision of the service. The information about the changes is input for the Capacity
Plan. Capacity Management can submit RFCs during the development of the plan.
Capacity Management supports distribution planning when the network is used for
automatic or manual distribution.
There is a close connection between the Capacity Database (CDB) and the CMDB. The
information provided by Configuration Management is essential for developing an
effective CDB.
Capacity Management advises Service Level Management about the feasibility of service
levels (for example response and cycle times). Capacity Management measures and
monitors performance levels and provides information for checking and where necessary
changing the agreed service levels and associated reports.
Capacity Management specifies the minimum capacity needed to continue the service in
the event of a disaster. The capacity needs of IT Service Continuity Management should
be constantly reviewed to ensure that they reflect day to day changes to the operating
environment.
Activities
The Capacity Management activities are described below for each sub-process.
Modeling
Modeling is a powerful Capacity Management tool and is used to forecast the behavior of
the infrastructure.
The tools available to Capacity Management range from estimating to extensive
prototype testing. The former is cheap and often adequate for routine activities. The latter
is usually only appropriate for large-scale implementation projects.
Between these two extremes, there are a number of techniques that are more accurate
than an estimate, and cheaper than an extensive pilot. In order of increasing cost they
include:
Trend analysis (cheapest)
Analytical modeling
Simulation
Baseline assessment (benchmark) (most accurate)
Application Sizing
Application Sizing considers the hardware needed to run new or changed applications,
such as applications under development or undergoing maintenance, or which may be
purchased at the request of the customer. These predictions include information about the
expected performance levels, necessary hardware, and costs.
This discipline is particularly relevant during the initial product development stages.
Clear information about the required hardware and other IT resources and expected costs
at this stage is valuable to management. This discipline also contributes to the drafting of
a new SLA.
These sub-processes include the same activities, however the emphasis is different.
Service Capacity Management addresses the provision of IT services, while Resource
Capacity Management addresses technology, required to deliver on those services.
Monitoring
Monitoring the infrastructure components aims to ensure that the agreed service levels
are achieved. Examples of resources to be monitored include CPU utilization, disk
utilization, network utilization, number of licenses, etc. (i.e. there are only ten free
licenses available).
Analysis
The monitoring data has to be analyzed. Trend analysis can be used to predict future
utilization. This may initiate efficiency improvements or the acquisition of additional IT
components. Activity analysis requires a thorough understanding of the overall
infrastructure and business processes.
Tuning
Tuning optimizes systems for the actual or expected workload on the basis of analyzed
and interpreted monitoring data.
Implementation
The objective of implementation is to introduce the changed or new capacity. If this
means a change, the implementation involves the Change Management process.
Demand Management
Demand Management aims to influence the demand for capacity. Demand Management
is about moving demand. A simple example: a user is running a poorly written SQL
report in the middle of the day, knocking other users out of the database and creating an
inordinate amount of traffic. The Capacity Manager suggests creating a job to run the
report overnight so the user has it on his desk in the morning.
Short-term demand management - where a recurring lack of capacity is
threatened in the near future, and where additional capacity is not easily available.
Long-term demand management - where the cost of upgrades cannot be
justified, although there are certain periods (e.g. between 10:00 AM and noon)
when there may be insufficient capacity.
Demand Management provides important inputs for drawing up, monitoring and possibly
adjusting both the Capacity Plan and the Service Level Agreements.
Demand Management can also involve differential charging (i.e. different charges at peak
and off-peak times) to influence customer behavior.
Process Reports
The management reports provided by the Capacity Management process include, on the
one hand, process control information in terms of Capacity Plan characteristics, resources
used to implement the process and the progress of improvement activities, and on the
other hand, exception reports about issues such as:
Discrepancies between the actual and planned capacity utilization
Trends in the discrepancies
Impact on service levels
Expected increase/decrease of the capacity utilization in the short term and long
term
Thresholds that, when reached, will require the acquisition of additional capacity.
Performance indicators
The success of Capacity Management is determined by the following key performance
indicators:
Predictability of the customer demand: identification of workload developments
and trends over time, and the accuracy of the Capacity Plan.
Technology: options for measuring the performance of all IT services, the pace of
implementing new technology, and the ability to continually achieve the agreements
laid down in the SLA, even when using older technology.
Costs
The costs of setting up Capacity Management must be estimated during the preparation.
The costs can be divided into:
Purchase of hardware and software tools such as monitoring tools, Capacity
Management Database (CDB), modeling tools for simulations and statistical
analysis, and reporting tools
Project management costs associated with the implementation of the process
Personnel, training and support costs
Facilities and services
Once the process has been set up, there are recurring costs of personnel, maintenance
contracts, etc.
Problems
Potential problems in Capacity Management include:
Unrealistic expectations - designers, management and customers often have
unrealistic expectations based on a lack of understanding of the technical
possibilities of applications, computer systems or networks. One of the tasks of
Capacity Management is to guide these expectations, for example by making
designers aware of the impact of their design (e.g. for a database) on the capacity
and performance. The effect of Capacity Management can also be overestimated,
particularly with respect to tuning the system and scheduling the workload. If the
operation of systems requires extensive tuning it is likely that the design of the
application or database is poor. In general, tuning cannot be used to obtain a higher
level of performance than the system was originally designed for. Most large
systems have scheduling algorithms that are generally more effective than
intervention by system managers. And of course, there is a cost associated with
tuning-it makes no sense for a highly paid engineer to garner a 3% performance
improvement after a weeks' worth of effort when a $100 stick of memory would
produce a 10% improvement. There is a larger cost to over-reliance on tuning-the
cost of managing systems which are not, within reason, 'plain vanilla'. Highly
'tweaked' parameters on different boxes, applications, or databases means
unintended consequences, and adds delays across all service management
processes, in maintenance, troubleshooting, etc.
Lack of appropriate information - it is often difficult to obtain required
information, for example for the Capacity Plan. It may be difficult to obtain reliable
information about the expected workload, as the plans of the customer are not
known or almost unknown, at least not in detail. This is also difficult for the
customer, as product life cycles are getting shorter and shorter. The only solution is
to make the best possible estimate and to update the estimate frequently when more
information becomes available.
Supplier input - if there is no historical data (for example when a new system is
purchased), Capacity Management often has to depend on the information provided
by suppliers. Suppliers normally use benchmarks to provide information about their
systems, but because of the major differences between testing methods it is often
difficult to compare information and it can be misleading about the actual
performance of the system.
Implementation in complex environments - implementation in complex
distributed environments is difficult because the magnitude of technical interfaces
creates a large number of performance dependencies.
Determining the appropriate level of monitoring - monitoring tools often provide
many options and may encourage investigations in excessive detail. When
purchasing and using these tools, it should be decided in advance at what level
monitoring should be carried out.