SOA Basics .......................................................................................................3
SOA Basic Structure .........................................................................................3
What SOA Means to Capacity Management .....................................................4
What Stays the Same ...................................................................................4
Capacity Management Challenges ...................................................................4
1 | TeamQuest IT Knowledge Exchange Series SOA and What it Means to Capacity Management
Service Oriented Architecture, or SOA for short, is probably the most discussed application
development topic today. Where and how it originated is a matter of opinion. Some say it was
initially devised by two Gartner Group analysts. Some say Sun Microsystems developed it to
support their JINI project. Some say it is an offshoot of Component Based Design or CBD and
others say it is a natural evolution from Structured Programming and Object Oriented Design.
Besides the variety of claims, there are a multitude of “SOA” solutions being offered, usually in a
form that touts a vendor’s own proprietary hardware and software solutions; actions somewhat
in conflict with SOA’s goal of transparent interoperability.
Standards organizations generally lag behind technologies and architectures, usually waiting
until general acceptance is clear before committing precious resources to set a standard. SOA
is no exception. Recently OASIS (Organization for the Advancement of Structured Information
Standards) issued a standard called “OASIS Reference Model for Service Oriented Architecture,”
however it provides basic definitions that are more suited to guiding vendors in developing
solutions than IT organizations implementing SOA. OASIS continues its work to better define
the different facets of SOA but we feel it will be some time before work can be completed and
Because of the wide variety of offerings and minimal standards, SOA means a lot of different
things to a lot of different people. In this paper, we will attempt to provide a simplistic view of
SOA based upon its goal of a reusable, interoperable, platform independent architecture, then
discuss how this new architecture impacts capacity management organizations.
2 | TeamQuest IT Knowledge Exchange Series SOA and What it Means to Capacity Management
SoA BASICS • Use IT infrastructure assets more
efficiently and effectively
determine new functions, changes to
existing applications, or sunsetting
It is important to remember that SOA, as applications no longer needed.
the name implies, is only an architecture. The architecture is comprised of
It is not a set of processes, procedures components, services, and processes. Application Architecture defines
or best practices; however those are Although the solutions differ depending business templates or scripts that
needed if you are to effectively and upon vendor products, there appears determine which services and/or
efficiently operate applications within to be a common theme thus we have components are invoked, in what order,
the architectural framework. chosen a 4-tier approach to help and when they are invoked. This area of
visualize SOA. 1 SOA is responsible for understanding
SOA provides a structure where changes in business processes, then
applications or application components • Component – An individual using the knowledge to modify or enhance
(i.e. services) more closely resemble executable module that performs a existing IT processes to satisfy them. The
business processes, where services are single, measurable task applications teams will integrate other
easily reused and interoperate well with existing services into the processes and,
each other. The architecture provides a • Service – A collection of components where gaps exist, develop or procure
framework by which services interface that provide a unique business new ones. When using SOA, applications
with one another and interoperate service such as an order processing teams become integrators more than
independent of how they are implemented service or a general ledger service. application coders.
and independent of whether they are The architecture permits services
internally or externally sourced. to be provisioned internally or
externally with equal ease and
When done correctly, additions and transparent to the service consumer. •HTTP
changes to the application portfolio can •SOAP
be accomplished quickly and easily, • Processes – A collection of
components and services that are
permitting the business to capitalize
invoked in a particular sequence in •XML
on opportunities while keeping IT
infrastructure costs low, improving order to mirror business processes. •J2EE
Common IT management reasons for Service Consumer Service Architecture addresses the
implementing SOA: structure of services and how they are
sourced. Some of the more common
• Develop architecture that can support solutions to support this architecture are:
and sustain shifting business
requirements Application Architecture Middleware is the technology and
protocols that permit the transparent
• Use technology-neutral components movement of data and information,
and services usually in increments called messages,
across the IT enterprise infrastructure.
• Speed time to market of new Service Architecture Middleware also permits seamless
applications and functionality integration of external services into
enhancements internal processes.
• Employ component reuse techniques Workflow utilities facilitate consistent
to reduce development times Component Architecture execution of business processes.
During construction of a workflow,
• Leverage reusable components discrete pieces of work, the order in
to reduce application inventories, which they are completed, and the
eliminate duplication of effort thus SoA BASIC people or business teams responsible
lowering ongoing support costs for completing them are identified.
StruCturE A script is developed in the workflow
• Increase transparency between utility. As work is assigned, the
Service Consumers are the people who
internally and externally sourced workflow utility manages the assigning
exploit IT services to perform business
components and services and completion of the individual steps.
functions. They are responsible for
using services as agreed upon in Service Where steps are not completed within
• Simplify interoperability across the specified time periods, the workflow
Level Agreements (SLAs). They usually
enterprise with a consistent approach utility invokes escalation procedures
pay for the service, either directly or
to application design
indirectly. They interface with IT to to manage the delays.
3 | TeamQuest IT Knowledge Exchange Series SOA and What it Means to Capacity Management
Converters, sometimes called mediators, application and business process views, Applications still consume server
may be used to normalize data which may include both internally and and network resources so the need
between diverse infrastructure, such as externally sourced components, are remains to gather performance data,
normalizing data between EBCDIC format needed to support management reporting store it for historical purposes, and
on IBM Mainframes and ASCII format on needs. Capacity plans must be determined generate management and progress
other platforms. In video format terms, looking across all affected infrastructure reports from it.
converters would permit European PAL components to ensure service quality
format to be displayed on North American across business processes. Business needs still need to be
NTSC devices or vice versa. understood and translated into
The need for modeling becomes more technology requirements. Those
Employing solutions such as those urgent because although there are requirements still are needed to
previously mentioned permits the fewer components to deal with, there feed existing planning processes,
application teams to seamlessly is greater service quality risk due interfacing with Finance to build, and
integrate components and services into to application interdependencies of then track financial plans supporting
scripts or flows that resemble business component sharing. IT infrastructure growth.
processes. When implemented correctly,
components and services can be added, More involvement with the business There is still a need to tune applications
changed or removed without any impact early in design cycles, as the reuse of to optimize the use of existing
to the service consumers short of any common components permits more infrastructure assets, so the need to
needed training. precise estimation of new application track trends and alert when potential
implementation and ongoing support problems may occur continue to exist.
costs. Using SOA’s modular approach, Bottlenecks still need to be identified
it is easier to predict new application before they negatively impact levels
performance. Since in many cases, of service.
components from existing inventory will
be used to satisfy requirements, capacity
management can extract performance
data and employ modeling tools to
accurately predict the impacts of the
additional usage on the IT infrastructure.
Reporting scope has to be expanded.
Because of the shared environments, The same application component or
detailedreportingmaynotbeasmeaningful service may be used across many
to business and IT management. A more business processes. Each process
meaningful approach will be to track the may generate work at different rates,
Component Architecture defines the performance of each business process depending upon needs. Therefore it may
make-up of components themselves from end-to-end or an enterprise view. be necessary to find ways to identify and
and the infrastructure upon which they segregate business process resource
run. Components usually perform a SOA requires more capacity usage internal to the components.
single, discrete function such as “get management involvement when
customer record” or “compute interest designing and scaling the enterprise For example, if we look at a customer
rate.” Components observe architecture architecture. Technology components service application, the day-to-day cross
protocols for sending and receiving data are just as important as the application section of customer types calling is
andinformationbetweenothercomponents development techniques. Quickly varied enough that you can accurately
and services. XML is a commonly employed developed applications are of little use if predict future growth using average
format for sharing data. the infrastructure cannot react to growth call metrics. Heavy query volumes
just as quickly. Therefore it is important could skew the average if you add a
to have capacity management and new service or process – using common
modules and services – that gives
WHAt SoA MEAnS infrastructure engineering involved in
SOA architecture plans and decisions. special assistance to good customers
to CAPACItY who have large amounts of purchase
history. If the different business
MAnAGEMEnt WHAT STAyS THE processes grow at different rates, it
could be difficult to accurately predict
WHAT CHANGES? SAME? future resource requirements. Therefore
it will be necessary to segregate usage
Capacity management processes and in the common components by business
Capacity management’s individual the work required to execute them process for more accurate predictions.
platform view is extended. End-to-end mostly remain the same. See the image on the following page.
4 | TeamQuest IT Knowledge Exchange Series SOA and What it Means to Capacity Management
management may need to devise ways
CUSTOMER SERVICE SPECIAL ASSISTANCE to associate reporting of IT events to
APPLICATION APPLICATION business durations.
SERVICE SERVICE SERVICE SERVICE
A B C D As you can see, capacity management
work to support SOA is not much different
from today. The same processes prevail,
Provisioning or sourcing common As an example, the legacy with a change in scope from individual
services may be difficult. When sharing a application could be converted to platforms to business processes.
particular service, you need to ensure web services. Although the legacy
piece would go away, some of the Since the viewpoint changes, new reports
the lowest common denominator underlying OS support overhead need to be created to supplement those in
is satisfied. That lowest common may not and some utility programs may place supporting existing platforms. Where
denominator may require substantially be used by other applications. external services are used, reporting needs
different levels of service than the to be put in place to ensure vendors deliver
more frequent users of the service. For Recovering all resources consumed on a expected levels of service and when they
example, a mapping service used by mainframe application is rare when it is do not, provide the data required to
customers to find stores may experience moved to a different platform. escalate and resolve the issues.
light usage but management prefers
customers to find store locations any Larger, more comprehensive testing Modeling becomes even more necessary
time of day or night. The warehouse environments would be required to test all to ensure service quality, plan for the
application uses the same mapping applications sharing common components future, and find and correct potential
service to optimize truck loading and because of interoperability complexity. bottlenecks prior to negative impacts to
delivery routes to the stores. Usage is This additional infrastructure could production service quality.
heavy but the warehouse only uses the substantially increase testing expenses.
service weekdays between 8 a.m. and As a result, capacity management may Testing rigor requires a larger scope,
5 p.m. As a result the common service be called upon to employ modeling requiring more testing resources.
could cost more since the lightest usage techniques to predict application Using shared components and services
requires more expensive service options. performance based upon results produced require test practices that span different
Therefore one must understand the in smaller test environments. business processes to ensure code and
trade-offs between lower development performance quality for all processes
costs by using a shared service and For outsourced services, capacity using shared components and services.
increased service costs by sourcing for management will need to determine More rigorous monitoring is needed and
the lowest common denominator. performance data requirements for modeling may provide lower cost options
inclusion in business process performance than large testing infrastructures to
During transition of legacy application reporting. Obtaining data can be difficult assure application performance.
functions to the new architecture, it if performance data accessibility is not
may be difficult to accurately plan addressed when the contracts are written. 1 Bieberstein, Norbert, Bose Sanjay,
capacity positions. In many cases, the Capacity management must be involved Fiammante, Marc, Jones, Keith & Shah, Rawn
applications run in shared environments in contract discussions to ensure needed (2006). Service-Oriented Architecture Compass
so it may not be clear as to what really data is available to ensure vendor service – Business Value, Planning and Enterprise
goes away in the legacy environments. levels are met and business processes Roadmap. Upper Saddle River: Pearson.
perform as expected.
Sprott, D, & Wilkes, L, (2004) Understanding
Planning could further be complicated if
Service-Oriented Architecture. Microsoft
legacy systems need to be available for Business processes may have durations Architect Journal. Retrieved August 2006
historical purposes, such as regulatory measured in hours or days whereas IT from http://msdn.microsoft.com/library/
reporting, for some undetermined process durations are usually measured default.asp?url=/library/en-us/dnmaj/
period of time. in seconds or minutes. Capacity html/aj1soa.asp.
If you like what you have read, click here for a
to the TeamQuest IT Knowledge Exchange Series.
5 | TeamQuest IT Knowledge Exchange Series SOA and What it Means to Capacity Management