Professional Documents
Culture Documents
Gary S. D. Farrow
Received (in revised form): 31st January, 2012
Triari Consulting, 30 Clothorn Road, Didsbury, Manchester, M20 6BP, UK.
Tel: +44 (0)161 445 5958; Fax: +44 (0)161 445 5958; e-mail: gary.farrow@triari.co.uk
Gary Farrow is Director of Triari Consulting, environment. Selecting the appropriate architec-
provider of systems integration and IT architec- tural style for payments processing is therefore
ture services for the financial sector. A lead an extremely important activity. This paper
architect on many successful, high-value proj- presents architectural patterns for payment sys-
ects, he has undertaken senior architect roles at tems integration. These enable the selection of
major banks and financial services firms in the optimal architectural solutions for payments
UK. Gary has broad expertise in large-scale sys- processing. Each pattern is described in terms of
tems integration, enterprise service bus archi- a number of generic Architectural Building
tectures and Service Oriented Architecture Blocks1 and their interactions.The general char-
(SOA), and deep domain specialism in payments acteristics of each pattern are compared, and the
systems. He has written many articles on SOA suitability of each pattern in supporting specific
and payments processing, discussing wide- business scenarios is presented. The paper also
ranging topics such as anti-patterns, the role of introduces the concept of the Payments
data and the optimisation of delivery models. Integration Space, which defines the scale of
Gary is a member of the IT Architecture the systems integration problem. This metric is
Certification Board for the Open Group. His pro- used to illustrate the role of the patterns in
fessional qualifications include Fellow IET, reducing complexity. It is shown that two spe-
Chartered Engineer and Open Group Master cific patterns, Payments Hub and Payments
Certified Architect. He holds BSc and PhD Service Bus, are candidates for the target end
degrees from Manchester University and is an state. Given significant differences in character-
alumnus of Warwick Business School, UK. istics of the two patterns, it is concluded that a
bank must make an early and overt choice
ABSTRACT about selecting the most appropriate target end
Regulatory, industry and competitive drivers state for payment systems integration.
dictate that payment initiatives within a bank
are many and continuous. However, a common Keywords: architectural patterns, trans-
situation is that the supporting IT systems action processing, Payments Hub,
have evolved in an ad hoc manner — there Payments service bus, systems integra-
are often limited separation of architectural con- tion, TOGAF
cerns, many tightly-coupled legacy systems and
duplication of processing capability. Systems
integration complexity is a limiting factor in INTRODUCTION
determining how quickly IT changes can be Deficiencies in payments systems process- Journal of Payments Strategy &
Systems
achieved. These factors detract from a bank’s ing manifest themselves in business terms Vol. 6, No. 1, 2012, pp. 15–36
Henry Stewart Publications,
ability to respond to changes in the business as: 1750–1806
Page 15
Farrow:JSC page.qxd 30/04/2012 13:49 Page 16
Page 16
Farrow:JSC page.qxd 30/04/2012 13:49 Page 17
Farrow
Finally, the adoption of each of the pat- bank staff to perform payments on behalf
terns as the candidate target end state for of a customer or for the bank itself. A
payments processing is explored. To this channel system will typically only initiate
extent, the degree to which each facilitates outbound payments. Communication from
the desired system processing objectives is the Channel components is usually by
assessed, focusing on the extent to which using an internal data format unique to the
the integration complexity is reduced. bank, although a derivation of an industry
format is also often the basis of this.
Page 17
Farrow:JSC page.qxd 30/04/2012 13:49 Page 18
Figure 1
Payments
processing
capabilities (after
Farrow)5
Page 18
Farrow:JSC page.qxd 30/04/2012 13:49 Page 19
Farrow
Figure 2
Payments
Processing
Spectrum
Page 19
Farrow:JSC page.qxd 30/04/2012 13:49 Page 20
Figure 3
Payments
Integration Space
Page 20
Farrow:JSC page.qxd 30/04/2012 13:49 Page 21
Farrow
Attribute Description
Payment Engine Routing Utility The sole function of the Payment Engine
Routing Utility is to interface to the
Overview Gateway and route payment instructions
This pattern represents a solution for inte- to and from the Payment Engines. Thus, of
grating a single Payment Gateway into the three categories, the Payment Engine
multiple payment engines or Product Routing Utility provides only Integration
Systems. Services. The combination of the Payment
The Payment Gateway is connected to Engine and the Product Systems provide
a single Payment Engine Routing Utility the rest of the payments processing capa-
(Figure 4). This, in turn, is connected to bility.
two or more Payment Engines/Product The Payment Engine Routing Utility
Systems. The Payment Engines and does not manage any business state. It may,
Page 21
Farrow:JSC page.qxd 30/04/2012 13:49 Page 22
Page 22
Farrow:JSC page.qxd 30/04/2012 13:49 Page 23
Farrow
Figure 5 Payment
Engine Routing
Utility Spectrum
Apportionment
Spectrum Apportionment
As per the Payment Engine Routing util-
ity, capabilities employed in this pattern are
middleware services, and hence the pat-
tern apportions capabilities from the left-
hand side of the Spectrum (Figure 7).
Capabilities from the Spectrum utilised
in this pattern include:
Page 23
Farrow:JSC page.qxd 30/04/2012 13:49 Page 24
Figure 7 Product
System Routing
Utility Spectrum
Apportionment
Page 24
Farrow:JSC page.qxd 30/04/2012 13:49 Page 25
Farrow
Spectrum Apportionment
This pattern construct provides middle- the payment instructions with bank-
ware services as per the two Routing specific technical or business data;
Utility patterns. It also provides the oppor- • Payments Repository, providing a ware-
tunity to introduce several payment house of all payments instructions for
Business Services (Figure 9). Considering audit purposes. In this pattern, the Front
the Capability Model, in addition to the End Payment Hubs has visibility of all
Routing, Transformation and Validation, inbound payments from the schemes
candidate Business Services include: and all outbound payments originated
from the channels. It is therefore a sen-
• Diary management, providing functional- sible point of control for updating or
ity for altering on files that are expected implementing a centralised Payments
to be received or sent to a scheme on a Repository;
given day according to the scheme • Intelligent payments routing, enabling
operating times; scheme selection based on general cri-
• Almanac, this being the determination teria, for outbound payments originated
of the Diary events for a given day; from the Customer Channels.
• Anti-money laundering (AML) and fraud
services, for example, sanctions checking Process services may therefore also be
on all inbound payments and event required to orchestrate these Business
feeds to AML and fraud management Services.
system components;
• Account validation, to validate the benefi- Scenarios
ciary account for an inbound payment; This pattern addresses the payments chan-
• Repair, supporting the correction of nel integration problem across the entire
payment instructions incorrectly bank. Thus, It is most relevant when the
formed from the channels; channel integration problem is more com-
• Enrichment, supporting enrichment of plex than the back-end integration prob-
Page 25
Farrow:JSC page.qxd 30/04/2012 13:49 Page 26
Figure 9 Front
End Hub Spectrum
Apportionment
lem. Circumstances where this may arise Back End Payments Hub
are when a bank is a member of many This pattern (Figure 10) is a generalisation
schemes and has many delivery channels of the Product System Routing Utility
but only a limited number of Product pattern. It also broadly equates to Back End
Systems. In such circumstances, this pat- Aggregation, this also being a further cate-
tern is considered a viable solution. gory of Payments Hub suggested by
Secondly, this pattern has been sug- Hayden et al.9 Again, the pattern represents
gested8 as an interim state towards achiev- an architectural formalisation of this pay-
ing a more complete payments integration ments processing strategy using the
solution (ie one that addresses front- and defined Architectural Model constructs.
back-end integration). In this case, the In this pattern, the Back End Payments
Front End Payments Hub supports a strat- Hub connects to one or more Product
egy for the phased introduction of Systems. It may also connect to some
common payments-processing services, Business Services, each providing a service
leading to the desired end-state solution. common to all payments processing.
The form of such end-state solutions is It is seen that the Back End Hub pat-
defined and discussed in two further candi- tern addresses the integration problem of
date patterns: Payments Hub and Payments integrating the Payment Engines to the
Service Bus. Product Systems. The Back End
Payments Hub provides the services to
Summary of characteristics integrate the Product Systems and other
components providing Business Services.
In this way it exposes Business Services
Front End Payments Hub
that may be used by several Payment
Engines to fulfil the payment process.
Attribute Value
Thus, in this pattern, the Payment Engine
Integration Capability Yes is the main provider of the process serv-
Business Capability Provides some shared ices, fulfilling a payments process using
services or delegates these Business Services exposed via the Back
to other components
Process Capability Some orchestration
End Payments Hub.
capability required
State Management Stateful or stateless, Spectrum Apportionment
depending on the specifics As per the Front End Hub, it may provide
of the Business Services a number of Business Services (Figure 11)
employed
or delegate these to other components
Connection Style Single Pass
providing shared Business Services.
Page 26
Farrow:JSC page.qxd 30/04/2012 13:49 Page 27
Farrow
Figure 11 Back
End Hub Spectrum
Apportionment
Page 27
Farrow:JSC page.qxd 30/04/2012 13:49 Page 28
Summary of characteristics
Attribute Value
Page 28
Farrow:JSC page.qxd 30/04/2012 13:49 Page 29
Farrow
Figure 13
Payments Hub
Pattern Spectrum
Apportionment
Page 29
Farrow:JSC page.qxd 30/04/2012 13:49 Page 30
Summary of characteristics
Payments Hub
Attribute Value
Description
The Payments Service Bus (PSB) is also a Business Service orchestration is pro-
widely used term for a generalised solu- vided solely by the Payment Engines, each
tion to payment processing within the of which may be tailored to provide the
bank (Figure 14). The differences between specific processing demanded by a scheme,
this and the Payments Hub are how clearly optionally complemented by the banks
articulated in terms of the Architectural own value-add services.
Model. Processing of an inbound payment
Unlike the Payments Hub, this pattern instruction, from receipt at a Gateway to
is premised on the existence of a number posting to an account in a Product
of separable Payment Engines, typically System, requires several passes through the
one per scheme. All integration to and PSB:
from the Payment Engines is provided by
the Payments Service Bus. In this way, a • collection from the Gateway with rout-
common utility is provided for all pay- ing and onward transmission to the cor-
ments integration within the bank. The rect Payment Engine;
Payments Service Bus can be considered as • calls to a number of the Business
a specialism of an Enterprise Service Bus, Services, orchestrated by the Payment
but specific to payments integration. Engine, but mediated by the PSB;
The PSB connects to all Consumer • selection of the recipient ledger system
Channels and Gateways. It also connects and associated account posting to the
to multiple Product Systems and to shared correct Product System.
Business Services. No payment instruction
passes between the payments system com- The PSB itself need only provide stateless
ponents without mediation by the PSB. ‘one shot’ services. It is the Payment
Page 30
Farrow:JSC page.qxd 30/04/2012 13:49 Page 31
Farrow
Figure 15
Payments Service
Bus Spectrum
Apportionment
Engines that manage any payment business Given the knowledge capital invested in
state if it is necessary to do so. the existing Payment Engines, likely com-
plexity of legacy implementations (imply-
Spectrum Apportionment ing huge effort to replicate and replace)
In this pattern, the Bus construct provides this pattern supports a strategy of their
only the integration services and hence continued use. Thus, the PSB solution
occupies the left-hand portion of the allows for their retention and focuses on
Spectrum. Multiple Payment Engines pro- addressing the integration problem
vide scheme-specific process services and between the Payment Engines and the
implement (or orchestrate) some or all the other payment solution components. In
Business Services, and hence are shown comparison with the Payments Hub pat-
occupying the middle and high end of the tern it therefore avoids significant effort in
Spectrum. As per the Payments Hub pat- Payment Engine replacement.
tern, the construct has visibility of all pay- This solution pattern was selected by a
ments and so Business Services across the large UK retail bank as a solution to the
entire Spectrum are applicable in this pat- payment integration problem. The bank
tern (Figure 15). was a full member of all UK and European
schemes with dedicated Payment Engines
Scenarios and having several major Product Systems.
As with the Payments Hub pattern, the The bank took a view, as highlighted
PSB also provides a generalised solution above, that consolidation of its diverse
to payments integration across the bank, range of Payment Engines into a single
solving the problems of front- and back- solution across the bank was a massive
end integration. This pattern does not undertaking, not feasible in a short to
require the replacement of legacy medium time frame. Rather a strategy of
Payment Engines; rather it recognises and leveraging existing Payment Engines, and
complements the scenario of multiple a policy of gradual, targeted, replacement
Payment Engines. was employed. The Payments Service Bus
The pattern is therefore useful in banks pattern was identified to support this strat-
where there exist multiple, diverse, egy. Technology selection and implemen-
deployed instances of Payment Engines. tation is ongoing within the bank.
Page 31
Farrow:JSC page.qxd 30/04/2012 13:49 Page 32
Page 32
Farrow:JSC page.qxd 30/04/2012 13:49 Page 33
Farrow
For simplicity, it is assumed that all Business Services does not detract from
Channels: the analysis given.
The original Payments Integration
• support payments types for all schemes; Space relating to this example is shown
and graphically in Figure 16a.
• a connection to all Product Systems is For the Payments Integration Space, an
required for account updates. overall complexity coefficient is intro-
duced, defined as the total sum of the inte-
In practice all payment types may not be grations. In the example defined:
offered from all Channels and some
account updates would be controlled by Example Integration Space Coefficient: 175
the Payment Engine. Such factors would
lower the integration complexity in real- The effect on the reduction in the
ity. This derives a specific table below Integration Space complexity is now illus-
(Table 3). trated for the two candidate end-states pat-
Note that the Integration Space does terns. By applying them to the example
not account for any integration to compo- problem space, the connectivity is re-evalu-
nents providing Business Services. These ated to give the revised Integration Space.
are excluded on the basis that:
Payments Hub Pattern Coefficient: 21
• There is duplication of functions spe-
cific to each transaction processing sce- Payments Service Bus Pattern Coefficient: 33
nario bundled into the application
components specific to the scenario. The Integration Space for each of the pat-
Hence, there are no connections to terns is shown visually in Figures 16b and
external services and hence no integra- 16c.
tion problem per se.
• It is unlikely that many common serv- Summary
ices exist pre-end state, and hence there Both Payments Hub and Payments Service
are no many-to-many connections to Bus are seen to provide a significant
resolve. reduction in Integration Space complex-
• If there are some shared services, then ity. However, they represent quite different
connection point is difficult to gener- solutions to the payments integration
alise in terms of the model presented. problem. The decision on choice of pat-
tern depends on several factors:
The net effect of excluding these is recog-
nition that the overall complexity is • the specific circumstances of the bank,
increased in all cases, so excluding the accounting for the uniqueness of each
Consumer Channel – – 15 30
Payment Gateway – – 15 5
Product System 15 15 9 18
Payment Engine 30 5 18 –
Page 33
Farrow:JSC page.qxd 30/04/2012 13:49 Page 34
Figure 16a–c
Complexity
reduction using the
Payments Hub and
PSB patterns
16a Original
Page 34
Farrow:JSC page.qxd 30/04/2012 13:49 Page 35
Farrow
Advantages Disadvantages
Payments Service Bus • Complements existing legacy Payment • Residual integration space remains
Engines; (does not require replace- more complex than that of the
ment of them). Payments Hub, albeit marginal
• Supports multiple technologies for • Difficult in practice to achieve pure
Payment Engine implementations integration capability as some business
• Enables ‘out of the box’ Payment decisioning (eg intelligent payments
Engine functionality to be leveraged routing) must take place outside of the
from vendors products Payment Engines
• Pure integration rather than • Resulting IT estate is more complex
replacement design and build effort with multiple product vendors relating
to each Payment Engine
Payments Hub • Least complex integration space, albeit • More development effort to reach
marginally end-state than the Bus as it requires
• Suits legacy replacement replacement and consolidation of each
• Single Architectural Building Block legacy Payment Engine.
for all payments scheme processing, • Single technology choice may not be
simplifying IT estate and supplier optimal for all scheme processing
management
• Single technology implementation and
simplified systems management
• Is an enabling architecture for custom
processing supporting uniqueness of
payments processing in each bank
bank and the general approach to pay- • separation of architectural concerns, with
ments processing within the business; components optimised for one purpose;
• the scale of the bank’s payments pro- • elimination of the tight coupling of sys-
cessing requirements; tems;
• the complexity of the existing IT archi- • provision of the foundation for shared
tecture for payments processing, com- payment services to be introduced.
prising the heritage Payment Engines
and Product Systems. A number of patterns for payment systems
integration have been presented and their
A summary of the advantages and disad- processing characteristics defined. The pat-
vantages of each of the two candidate pat- terns are shown to relate to specific appor-
terns is presented in Table 4. tionments of the Payments Processing
Spectrum, which itself provides a useful
visual tool for comparing them.
CONCLUSION IT scenarios to which the patterns are
Payments integration complexity is a relevant have also been highlighted and
major barrier to enabling the business discussed.
agility of a bank. When the integration Certain patterns are seen to address the
problem is addressed, desirable architec- front-end (scheme and channel) integra-
tural goals are achieved, which can tion problem, while others address the
improve business agility: back-end (Product System) integration
Page 35
Farrow:JSC page.qxd 30/04/2012 13:49 Page 36
problem. The relevance of employing spe- ing modernisation. Such planning activity
cific payment Business Services in each of is anything but trivial and must account
the pattern contexts has been demon- for many factors, not least the demanding
strated. time constraints imposed by regulatory
Of the six patterns identified, only deadlines, which tend to dictate non-
Payments Hub and Payments Service Bus strategic solutions. An iterative and incre-
patterns offer a holistic solution to this mental transition strategy can help
payments integration problem, addressing accommodate such factors and a roadmap
both the front-end and back-end integra- should be determined on this basis.
tion problems. For this reason, the
Payments Hub and Payments Service Bus
ACKNOWLEDGMENTS
patterns are presented as prime candidates
for the target end state for payment sys- Thanks are due to Greg Brougham for his
extensive review comments and suggestions for
tems processing. The architectural differ-
improvements to the architecture model.
ence between a Payments Hub solution
and a Payments Service Bus solution has
been qualified using the respective pat- REFERENCES
terns. Further, the impact of the patterns (1) The Open Group Architecture
on the overall integration complexity has Framework (2009) TOGAF 9, ISBN
been quantified in terms of the reduction 978-90-8753-230-7, available at:
in Integration Space complexity. http://www.opengroup.org/togaf
There are considered to be significant (accessed 2nd March, 2012).
implications in choosing one or other of (2) Farrow, G. (2011) ‘The Payments Hub
the end state patterns. In particular, the spectrum; A model for the design of
technology selection for the respective Payments Hubs’, Journal of Payments
integration constructs is influenced heavily Strategy & Systems, Vol. 5, No. 1,
by the characteristics identified in the pat- pp. 52–72.
(3) TOGAF 9, ref. 1 above.
terns. Depending on choice of end-state
(4) Farrow, ref. 2 above.
pattern, long-term commitment to retain (5) Ibid.
or decommission specific legacy technolo- (6) Ibid.
gies may be inferred, and alignment to (7) Hayden, R. Akash, L, Ledford, S. and
overall IT strategy must be assessed and Nunn, C. (2010) ‘Payments Hubs:
validated. Consideration must also be Redefining the industry’s infrastructure’,
given to how development of a payments McKinsey Quarterly, June, available at:
processing system is undertaken within the http://www.mckinsey.com/clientservice
bank; how the development teams are /Financial_Services/Knowledge_
structured; and the long-term skills Highlights/Recent_Reports/
required. Payments.aspx (accessed 16th February,
In conclusion, given the significant 2012).
(8) Ibid.
implications of pattern selection, a bank
(9) Ibid.
must take an early and overt decision as to (10) Farrow, G. (2011) ‘The Canonical Data
the desired target end state for payments Zone: Issues in data selection for Service
processing. Once this decision has been Oriented Architectures’, The Open
reached, business events on the planning Group Conference, 9th–13th May,
horizon can be assessed in terms of their Westminster, London.
suitability as triggers for payments process- (11) Hayden et al., ref. 7 above.
Page 36