How to Test the Internet of Everything

How to Test theInternet of Everything

@paul_g

erra

rdPaul [email protected]

gerrardconsulting.comSlide 1

Agenda

• The Testing Challenge• The Scope of IoE Testing• Test Strategy for the IoE• The Future for Testing and Testers• Close

Slide 2Intelligent Definition and Assurance

Background

• The IoE will dominate our industry for 10 to 20 years

• Right now, it is very confusing• Standards emerging, but current applications

are bleeding edge, speculative or exploratory• Security and privacy risks dominate today• But functional and integration complexity

challenges must also be overcome• There are also broader societal risks• Government, companies and individuals need to

pay attention.


Introduction

• IoE will be 'everything on the internet' soon

• A strategy for 'Testing Everything'?– Dimensions, influences, and opportunities– Can only paint a picture– Suggest what testers need to think about

• The challenge we face is quite different from what we have tested before.


Scope of the Internet of Everything• The scope of IoE is growing and includes:

– Vast number of static devices, mostly sensors– Mobile devices: cars, buses, trains, planes, ships,

satellites...

• It will include:– Objects in homes, cities, work, hospitals, social venues– Wearable, human-embedded devices, phones and tablets

• IoE integrates with 100s of services, payment and legacy systems

• Eventually, IoE will be called the Internet• The test strategy for the IoE is the test

strategy for everything.


The Biggest Thing in 30 Years?• Continuous delivery and DevOps bring

factory automation processes into software• After Waterfall and Agile, CD and DevOps

become 'the third way'• The IoE will affect everyone on the planet.• From 50-700bn devices over the next 5 to 20

years; no one knows• We are on a journey that will increase the

scale of the internet by one hundred times• How will we test it?


The Testing Journey(from the 1980s to the present)

• Screen based applications running on dumb terminals – 1980s

• GUI applications and Client/Server – 1990s• The emergence of the Internet - 1995• 'Mobile Revolution' - today• Internet of Everything - emerging

– Smart phones, buildings, cars, ships, planes and cities– BigData, pervasive, wearable computing– Massive scale

• IoE brings new levels of complexity and scale• Non-functional risks are well-understood – what's new?• Functional testing and simulation at very large

scale.


If CMOs hold the budget…

• There is a shift in power in the mobile space• If marketers hold the money for building IoE:– IT will have to meet short and long term business

goals– Development must deliver, experiment and learn

• Market testing and experimentation to drive rapid evolution

• DevOps, Continuous Delivery, high levels of automation and simulation

• Test analytics blended with production analytics.


Connecting the Unconnected• 'Internet.org by Facebook' wants to provide cheaper

access to data by (to perhaps 1/100th of the current price)

• Developing economies will probably lead the way in implementing the IoE with mobile internet

• But what networking technology/protocols will be used for local connectivity?

• "No technology dominates yet. We are waiting for technologies to improve in range and performance. The standard technology for connecting the IoT hasn't been invented yet."

• We are still at the start of the IoE journey


The Scope of IoE Testing

Functional testing

• Hardware-Level functionality– Most of this testing will be performed by

manufacturers

• Object and Server level functionality– We'll have to test simple web-apps to systems with

many complex sub-systems

• Integration and Interaction at all levels– Companies, testers will have to use judgement on

deciding what to test and how much

• Collaboration confusion– Use expectations and user testing of collaborating

systems will be difficult.


Testing @ Scale

• Scalability–We'll have to do some large-scale load,

performance and stress testing

• Wearables and embedded–We will need to simulate and test

thousands of devices carried by or embedded in people

• Everything connected–We'll need to do location-based tests at

very large scale.


Networks

• Moving networks– We'll have to test security and resilience of our systems

against alien or rogue networks

• Network security risks at multiple levels– Networks will need to be hardened and tested at all levels,

not just 'perimeter'

• Device registration, provisioning, failure and security– Power-down, power-up and automated authentication,

configuration and registration processes will need to be tested

• Mobile objects– Power, interference, network strength, roaming and

jamming issues will all need testing.


Big Data

• Logistics– A substantial data storage service will have

to be tested (performance, backup/recovery etc.)

• Analysis and visualisation– These outputs and processes will need to be

tested.

• Personal and corporate privacy– Testing will need to embrace the ethical use

of data and it's protection from misuse.


The Scope of IoE Testing

• Not all IoE systems will be huge or complex• But we face a wider range of concerns than

ever before:–Most bring a new technology and risk profile– The range of testing issues we need to consider– The scale of the testing required has increased

• The approaches we will need to tackle the risks of failure will have to change.


Test Strategy for the IoE

Modelling, Testing and Test Data• Tools to model the IoE are only just emerging• We'll need to use Big Data test techniques• 'Pattern-based test design'– Generate tests from patterns to simulate scenarios– Oracles will derive from conventions or common-

sense– Personas as patterns of user or system behaviour– Goal: to feed automated tools large numbers of

scenarios

• Developers will need to make systems testable through High Volume Automated Testing.


Test Environments; Testing in the Field• Test environments will vary widely

– A test lab for a home could be set up in any office– An urban environmental management system will be

harder to create

• Scale of field testing will vary widely, but will become a common (and possibly mandatory) testing stage

• Barcelona has set up an urban-scale testing facility– Barcelona Urban Lab is positioned as a field test

environment to prove the viability of innovative systems.

• Expect other cities around the world to offer similar services.


Tool Support

• Automated support will be essential and probably dominate testing

• Most components tested through APIs, web or other services

• New protocols emerging; custom drivers may be required

• The challenge will be, "how do we design the thousands or millions of tests to feed the tools?"


Test Analytics

• DevOps processes and production analytics monitor system vital signs like sensors– These analytics are essentially the same

• Production systems are platforms for experimentation– The code that serves marketers can also

serve developers and testers

• Analytics code is your sensor network• DevOps processes are 'things', too.


Performance Testing and Test Data• Local network protocols

– IP based (XMPP, MQTT); few tools, but cloud services emerging

– Low Power/Lossy Network (ZigBee, 6LowPAN, DASH7/RFID, Bluetooth and NFC); only need simple drivers to 'chirp'

• Coherent data may be a problem, e.g. location data– Cars must follow the road network– People can only walk through certain places in a shopping

mall– So you will have to capture the GPS data using a tool

perhaps

• To create unique paths, you could edit together segments of intersecting paths.


The Future for Testing and Testers

We need to change the way we think about testing

New Model for Testing

"We explore sources of knowledge to build test models that inform our testing"

• Forget logistics; focus on 10 thinking skills• Not a process with sequence,

inputs/outputs, entry/exit criteria• What is the model useful for?– Separates logistics from the core testing skills– Clarifies two modes: exploration and testing– Supports all testing, regardless of methodology,

manual v automated, structured or agile.

Intelligent Definition and Assurance Slide 23

New Model Testing

29 page paper: http://dev.sp.qa/download/newModelSlide 24

http://dev.sp.qa/download/newModel



Manual v Automated Testing?• Automation will not make testing easy; it will

make testing possible• More and more, DevOps processes are a

source of data for analysis just like production• You might not need to be a programmer but

you may need to:– Write your own utilities, download, configure and

adapt open source tools or create automated tests– Have informed conversations with developers

• Should testers learn how to write code? Yes!


Shift-Left

• Every company seems to be 'shifting left'– Developers responsible for testing – no testers– Testers embedded in development teams– Testers morphing into business or systems analysts

• Testing thinking is being moved left and earlier in the process

• The New Model identifies these thinking activities

• Testers need to learn how to 'let go' of testing, so view this as an opportunity, rather than a threat.


Close

• Testing the IoE is the next level of testing in terms of complexity and scale

• The tools we need don't yet exist– Test models, test data generators and automatic

oracles– Simulation/High Volume Test Automation tools

• Testing in the lab and experimentation in production will become one discipline – test analytics

• Some new test approaches will be required• The New Model is an attempt to trigger new

thinking in this area.


How to Test theInternet of Everything

@paul_g

erra

rdPaul [email protected]

gerrardconsulting.comSlide 28

Education

How to Test the Internet of Everything