The tale of three ATAMs ... Dr. Andrzej J. Knafel Stage & - - PowerPoint PPT Presentation

the tale of three atams
SMART_READER_LITE
LIVE PREVIEW

The tale of three ATAMs ... Dr. Andrzej J. Knafel Stage & - - PowerPoint PPT Presentation

The tale of three ATAMs ... Dr. Andrzej J. Knafel Stage & Prologue: Scenery, Actors, Use Cases, History Act 1: ATAM of the "Old System" Act 2: ATAM of the "Reference Architecture" Act 3: ATAM of the "Implementation


slide-1
SLIDE 1

The tale of three ATAMs ...

  • Dr. Andrzej J. Knafel
slide-2
SLIDE 2

Stage & Prologue: Scenery, Actors, Use Cases, History Act 1: ATAM of the "Old System" Act 2: ATAM of the "Reference Architecture" Act 3: ATAM of the "Implementation Architecture" Epilogue: Conclusions

2

slide-3
SLIDE 3

The stage …

Scenery

Business applications at Roche enterprise as sources or destinations of data and interactions … … connect through Roche intranet, services, Internet and customer ´s IT with … … Roche medical systems (devices) located at clinical laboratories worldwide.

3

slide-4
SLIDE 4

The stage …

Actors: Roche Devices for In-Vitro Diagnostics

Person icon made by Freepik from www.flaticon.com 4

slide-5
SLIDE 5

The stage …

Actors characteristics

1. Device locations – Worldwide

  • Different medical device regulations
  • Different privacy and security regulations

– Various maturity of IT environment

  • High, e.g., hospital clinical labs
  • Low, e.g., pharmacies, doctor practices

2. Connectivity – Number of connected devices: n x 106 – Data volume exchanged:

  • MBs/day for ~9/10 of devices
  • GBs/day for ~1/10 of devices

5

slide-6
SLIDE 6

The stage …

Major Use Cases

  • 1. Registration and configuration of devices
  • 2. Download

Transfer of data from Roche enterprise applications to the devices

  • 1. Processing parameters
  • 2. Software updates / upgrades
  • 3. Upload

Transfer of data from devices to Roche enterprise applications

  • 1. Alerts / Events
  • 2. Operational counters
  • 3. Logs
  • 4. Raw measurements
  • 4. Remote interaction, e.g., screen sharing

Gateway Device Business Applications End-to-End Processes Remote Service Remote Service Apps

⇔ Registration & configuration ← Download data & SW → Upload data ⇔ Remote interaction 6

slide-7
SLIDE 7

Epilogue Act 3 Act 2 Act 1 Prologue

The play in 3 acts …

Plot: the journey of architecture decision making

ATAM 1 of

  • ld system

+ AIW Improvement of Current Systems “Homegrown”

C

“Commercial”

D

Elaboration of Reference Architecture ATAM 2 of “Reference Architecture” Pre-Selection of Solution and Service Providers ATAM 3 of “Implementation Architecture” Proof-of-Concept Implementation start Months N N+4 N+16 “Homegrown” “Commercial” ATAM – Architecture Tradeoff Analysis Method AIW – Architecture Improvement Workshop

7

slide-8
SLIDE 8

The prologue …

History

Homegrown (2000)

  • Incrementally developed at

Roche until 2007

  • Still operational under load

>7x original specifications

Commercial (2008)

  • Bought in 2008
  • Unsuccessful roll-out until

2014

Future demand

  • High increase in number
  • f connected systems
  • New use cases

increase load of data

  • More demanding Quality

Attribute measures

8

slide-9
SLIDE 9

Stage & Prologue: Scenery, Actors, Use Cases, History Act 1: ATAM of the "Old System" Act 2: ATAM of the "Reference Architecture" Act 3: ATAM of the "Implementation Architecture" Epilogue: Conclusions

9

slide-10
SLIDE 10

Act 1 … where are we?

ATAM1 - risks with the existing systems

# Quality Attributes Risks in System 1 Availability Commercial 2 Reliability Commercial 3 Accuracy 4 Performance / Scalability Commercial Homegrown 5 Interoperability 6 Extensibility / Modifiability Homegrown 7 Maintainability 8 Security

Concern of not being able to fulfill the growing business needs and customer expectations

70 Risks

Evaluated 32 scenarios

10

slide-11
SLIDE 11

Act 1 … where can we go?

AIW – planning improvements

  • Outcome summary
  • rganization change recommended
  • improvements to existing systems

architecture

  • looking for alternative technical solutions
  • Decision to use ATAM in further phases of the project

11

slide-12
SLIDE 12

Stage & Prologue: Scenery, Actors, Use Cases, History Act 1: ATAM of the "Old System" Act 2: ATAM of the "Reference Architecture" Act 3: ATAM of the "Implementation Architecture" Epilogue: Conclusions

12

slide-13
SLIDE 13

Act 2 … examining the options

Improvements

  • Successful

– The old “Homegrown” system can handle load ~18x higher than specification – Expanded time window for finding good solution

  • Unfortunate

– Not resolved risks and limitations

  • f the “Commercial” system
  • High importance of searching for alternative solution

13

slide-14
SLIDE 14

Act 2 … looking for alternative

Reference Architecture

Business Applications Integration Interfaces to Instrument Applications Cross-cutting Concerns, e.g., Fault handling, Authenticity, Integrity, …

Gateway Device Business Applications End-to-End Processes Remote Service Remote Service Apps

⇔ Registration & configuration ← Download data & SW → Upload data ⇔ Remote interaction

  • 1. Reference Architecture

– Patterns solving all major risks from ATAM1 – Technology agnostic – Used in solution & vendors evaluation (part of RFP)

  • 2. Architecture Governance

– Organization – Strict governance of

  • Interfaces testable for

conformance

  • Cross-cutting concerns
  • Fault handling built into

end-to-end xfer protocol

  • Command –

Response behavior with sender-controlled, time-constrained, multi-level acknowledgment

  • Service Interface

Gateway - Devices

  • Multiple, extensible

channels

  • Multiple deployment

models for connectivity 14

slide-15
SLIDE 15

Act 2 … where are we?

ATAM2 - reference architecture & 1st solution

# Quality Attributes Risks 1 Availability

  • k

2 Reliability

  • k

3 Accuracy

  • k

4 Performance / Scalability

  • k

5 Interoperability

  • k

6 Extensibility / Modifiability

  • k

7 Maintainability

  • k

8 Security

  • k

Concern of many unknowns, level of architecture decisions, and management of the service / solution provider

23 Risks

Evaluated 38 scenarios

15

slide-16
SLIDE 16

Act 2 … where should we go?

Deriving planning decisions from ATAM2

  • 1. Stabilize the Reference Architecture & specify interfaces in detail
  • 2. Define a Proof-of-Concept to confirm architecture decisions
  • 3. Continue with service & solution provider selection (2nd phase RFP)

– Define “required” SLAs for service provider based on the Quality Attributes – Distribute to pre-selected candidates – Quality Attributes (incl. scenarios) – Reference Architecture – Required SLAs – Commercial requirements (responsibility outside of Architecture dept.) – Assess implementation architecture based on pre-defined evaluation criteria

16

slide-17
SLIDE 17

Stage & Prologue: Scenery, Actors, Use Cases, History Act 1: ATAM of the "Old System" Act 2: ATAM of the "Reference Architecture" Act 3: ATAM of the "Implementation Architecture" Epilogue: Conclusions

17

slide-18
SLIDE 18

Act 3 … architecting examples

Cloud related Remote Service requirements

1. Connection performance of the “last mile” to the customer latency based connection node location 2. Exchange of data with Roche Business Applications deployed – in the same cloud – in the different cloud –

  • n premises

3. SLAs based on Remote Service use cases and not individual services of cloud provider 4. Global availability – conformance to regional & country regulations

18

slide-19
SLIDE 19

Act 3 … architecting examples

Remote Service for the IoT Cloud

The Availability, Reliability and Performance of connectivity and data processing are well supported by Cloud technology, if the architecture properly utilizes its capabilities

Apply best cloud practices

  • Asynchronous Messaging
  • Compute Partitioning
  • Autoscaling
  • Data Partitioning
  • Data Replication & Synchronization
  • Multiple Datacenter Deployment
  • Service Metering
  • Caching

Use cloud design patterns

  • Compensating Transaction
  • Health Endpoint Monitoring
  • Messaging
  • Scheduler Agent Supervisor
  • Retry
  • Queue-based Load Leveling
  • Event Sourcing
  • Competing Consumers
  • Sharding

19

slide-20
SLIDE 20

Act 3 … architecting examples

Complex Events Processing (CEP)

  • Connectivity Heartbeat Monitor

– Sequence of events to detect anomalies

  • Connectivity Response Monitor

– Multi-level commands-response tracking – Time-out tracking

  • System / Device Health Monitoring

– Real-time security breach detection – Data and identity protection services – Degradation of customer experience

  • Preventive Maintenance

– Multi-alert pattern analysis – Automatic service scheduling

  • Benchmarking

– Notification when diagnostics process is degraded – On-line calculation of statistical parameters

20

slide-21
SLIDE 21

Act 3 … where are we?

Proof-of-Concept

Setup 1. Devices (n x 105) simulated by VMs in one cloud environment 2. Prototype of Remote Service in another cloud environment (different technology, different provider) 3. Evaluation of data logs with measurements Results 1. All scenarios fulfilled the required Quality Attributes 2. Some results exceeded the expectations

21

slide-22
SLIDE 22

Act 3 … where are we?

ATAM3 – target implementation architecture

# Quality Attributes Risks 1 Availability

  • k

2 Reliability

  • k

3 Accuracy

  • k

4 Performance / Scalability

  • k

5 Interoperability

  • k

6 Extensibility / Modifiability

  • k

7 Maintainability

  • k

8 Security

  • k

Concern of integration of the Remote Service into Roche business processes, incomplete operational concepts, migration.

23 Risks

Evaluated 64 scenarios

22

slide-23
SLIDE 23

Stage & Prologue: Scenery, Actors, Use Cases, History Act 1: ATAM of the "Old System" Act 2: ATAM of the "Reference Architecture" Act 3: ATAM of the "Implementation Architecture" Epilogue: Conclusions

23

slide-24
SLIDE 24

Epilogue

Conclusions

  • 1. Risks characteristics changed with improving the architecture

– ATAM1 (“Old System”): Serious technical deficiencies and organizational issues – ATAM2 (“Reference Architecture”): Many unknowns and vendor management – ATAM 3 (“Implementation target architecture”): Migration and Operations

  • 2. Each ATAM review resulted in decisions and planning leading to the next ATAM
  • 3. ATAM reviews started being an integral part of system engineering
  • 4. ATAM reviews are driving targeted organizational and technical changes

24

slide-25
SLIDE 25

The journey of architecture decision making …

ATAM 1 of

  • ld system

+ AIW Improvement of Current Systems “Homegrown”

C

“Commercial”

D

Elaboration of Reference Architecture ATAM 2 of “Reference Architecture” Pre-Selection of Solution and Service Providers ATAM 3 of “Implementation Architecture” Proof-of-Concept Implementation start Months N N+4 N+16 “Homegrown” “Commercial”

ATAM used as evaluation of system development & organizational change

25

slide-26
SLIDE 26

Doing now what patients need next

26