4. Nationwide Building Society
“How Nationwide has become more efficient over the SDLC by
implementing CA Test Data Manager & CA Service Virtualization”
6. Nationwide Building Society
• UK only based Financial Services organisation
• Roots trace back to 1846
• Has grown to be the worlds biggest Building Society, through Mergers &
Acquisitions
• Nationwide has a relationship with 1 in 4 households in the UK
• 17,000+ employees
• Customer Channels include, but not limited to
– 700+ Branches
– Web, Mobile, Tablet, Smartwatch
6
7. “DaveOps” - A centralised capability,
federated into Projects and Dev Teams
7
Automation
Data
Virtual Env.
(Service
Virtualisation)
Test
Tooling
Performance
Security
Test
Operations
Test
“DaveOps”
Mortgages Savings Core
Banking
Operational
functions (HR
etc)
And and
and…..
Digital
Channels
The Four Pillars of
DaveOps
• Invest in people
• Collaborate and
celebrate
• Don’t be part of
the problem, be
part of the
solution
• Don’t reinvent the
wheel, make it
rounder
8. Some Stats
• IT Development is broken into Projects and Development Centres
• There are around 130 projects of various sizes in progress at any point in time
• The majority of IT change within these projects will happen within 13 Development Centres
• Technologies include, but not limited to Mainframe, Off the shelf /w customisations, Middleware
SOA / API, Mobile
• ~1,000 resources in Test including strategic partnerships
• DaveOps has a team size of ~90 FTE
Rough alignments of team size, taking into account a shared resource pool
– Data ~15 FTE delivery up to 50 work packets p/month
– Service Virtualisation ~5FTE adding to & maintaining a catalogue of ~400 Virtual Services
8
9. Test Data in NBS – in the beginning
• Why did we start?
– The Data Protection Act (DPA)
– Wasted time and defects associated with data
• Where did we start?
– Customer masking programme for Test
• Purchase of CA Test Data Manager (c2010)
• Creating a centralised capability for providing data to Test teams
9
10. Where are we now – what did we learn
10
Wholly centralised doesn’t work – you need to understand what the
Test team want
A centralised focus is important to maximise reuse and
consistency – a factory of data creators is really powerful,
they will maximise use of CA Test Data Manager which most
will never see
Containerisation of data will make or break an effective data
team – the 80/20 rule of data is absolutely true…. People always ask
for a variation of the same thing – CA Test Data Manager excels at
this through “Cubes”
Don’t get stuck in the weeds when structuring how to
deliver data at the enterprise – look at the organisations
data store at an abstract level… there aren’t that many
Data is so much more than a good sql statement engagement is
key with Testers, Build, Design & make friends with IT Security, they
set the rules for maskingYou cant build a Test Data capability in isolation the next
time you use this capability, the system will have changed
Getting teams to self service is an achievable nirvana of
Test Data in continuous delivery dev teams TDOD in
digital testing sprints
Apply good practices from other areas and share resources if
you can – Automation, SV and Performance is especially good for
this – Test Matching is a real example
Your team will become extremely knowledgeable across the SDLC
and can influence up the value chain – DAVE are leading initiatives
to drive efficiencies' using models and coverage techniques, it’s not a
coincidence Agile designer and CA Test Data Manager are in the same
family….. My Test scenarios, are my Data requirements, are my
Coverage measure
11. The Test Data Service Operating Model
11
The Test Data
Warehouse
The Test Data
Factory
The Dev Team
The Test Data
Service
The Dev Team
The Dev Team The Test
Data
“Shapers”
12. The Test Data Service – what we do in TDM
12
Production Data Source
Data
Sampling /
Profiling
Subsetting
Data
Obfuscation
ETL/EDI/Flat
Files
Generated data
Subset
Masked
Created
Test Data Source
Test
Application
Test
Automation /
SOA
The DPS Data Provisioning Warehouse
Mass Edits Data Bulking
Data
Explosion
Data
Inheritance
Test Matching
13. Service Virtualisation in NBS – in the beginning
• Why did we start?
– Highly coupled infrastructure
– Challenges around getting dedicated “real tin” at the right time
• Where did we start
– Programme to introduce next generation online banking for
Nationwide
• Purchase of CA Service Virtualization (c2010)
• Focused on Performance Testing specifically
• Creating a centralised team within the programme
and migrating centrally into BAU
13
14. Where are we now – what did we learn
14
Testers and Developers speak a different language a
successful SV team can speak both & importantly translate
Service Virtualisation defects don’t actually exist a big mindset
shift that an VS is only as good as the requirements the Dev / Test
team have for it
Engage early to understand the Test Strategy you cant
replace things that need Testing. Importantly Test approach
needs to segregate out what's changing and what isn't
The “Transfer Problem” will quickly become apparent if it
exists in your Test Teams. A good VS will encompass coverage,
for this we need to apply coverage and design techniques
Don’t re-invent the wheel - Centralised, reusable assets
in catalogues create agility afterall we are replicating and
interacting with a middleware layer that is there to do exactly
the same thing
Is there formal SOA governance in the org? if there is, piggy
back it. It’s the fast track to so much you’d what to do with SV and
a clue to where the issue lie which SV can mitigate
Federate the SV servers recent versions of SV will allow us
to dedicate virtual reference servers into Dev / functional test
teams for agility.
Data and SV are fundamental to achieving robust, repeatable
automation so is a good testing model underpinning it…..
A centralised reporting & governance capability is key in
dispersed teams
Mainframe can also be virtualised much of our IT change
doesn’t touch mainframe