The speakers will start by going through the pitfalls of microservices and distributed systems, including how you can troubleshoot and solve typical failures. Next, they will present our Node.js toolset including custom networking modules, Hystrix/Eureka/Turbine integration, and the Mozzle/Riemann/Influx/Grafana monitoring stack that help us build, debug, deploy, and monitor their application at scale on Cloud Foundry. Finally, they will outline how neural networks and the new container networking feature could help with operating such applications.

1. Distributed Cloud Native
Applications with Node.js
Resistance Is Futile, Resilience Is Crucial

2. Hristo Iliev & Stoyan Rachev
Abacus committers
hello!
2

3. Distributed Cloud
Native Applications
with Node.js
Resistance Is Futile, Resilience Is
Crucial

4. 1.
CF Abacus
Architecture 101

5. COMPUTE STORAGE SERVICES
CF Abacus Billing Engine Invoice
Metering APIs
App Usage
Monetizing Your Applications Using Cloud Foundry Abacus and Billing Engines - Pankaj Kumar, SAP

6. COMPUTE STORAGE SERVICES
CF Abacus Billing Engine Invoice
Metering APIs
App Usage
Monetizing Your Applications Using Cloud Foundry Abacus and Billing Engines - Pankaj Kumar, SAP

7. Usage event
Collector
receives and
stores usage
Meter
applies
metering
formulas
Accumulator
accumulates
daily and
monthly usage
Aggregator
aggregates by
org, space,
service etc
Reporting
provides usage
summaries
collector
DB
meter
DB
accumulator
DB
aggregator
DB
Usage
Docs

8. 2.
Network
Pitfalls & recipes

9. “Fallacies of distributed computing:
✘ The network is reliable
✘ Latency is zero
✘ Bandwidth is infinite
✘ Topology doesn't change
https://blogs.oracle.com/jag/resource/Fallacies.html

10. Network Problems
Unreliable network
Insufficient bandwidth / latency
Application availability
✘ Crashes
✘ Updates

11. Fault Tolerance
Retry
✘ Seriously?!

12. Fault Tolerance
Retry
✘ Seriously?!
✘ Yes: with exponential backoff

13. Fault Tolerance
Retry
✘ Seriously?!
✘ Yes: with exponential backoff
Circuit Breakers
✘ Hystrix style

14. Bandwidth & Latency
Batching
✘ ↓Connections & latency
✘ ↑Throughput

15. Bandwidth & Latency
Batching
✘ ↓Connections & latency
✘ ↑Throughput
Throttling
✘ Limit requests

16. Availability
Platform
✘ Updates
Application
✘ Restarts
✘ Updates
Replay

17. Scaling
Availability → Scaling
✘ Cost
✘ Discovery

18. Monitoring & Discovery
Discovery
Cell
App 1 App 2
App 3 App 1
App 2
Router

19. Monitoring & Discovery
Discovery
Instance access?
Cell
App 3
Router

20. Monitoring & Discovery
Discovery
Indexed apps
✘ “collector-0”
✘ Single instance

21. Monitoring & Discovery
Performance? Security?
Discovery
Indexed apps
✘ “collector-0”
✘ Single instance
App 1 App 2
Router /
LB

22. Container networking
Direct container-to-container
communication

23. Indexed apps
✘ “collector-0”
✘ Single instance
App-to-App
✘ Via router / LB
Container Networking
Monitoring
Discovery
Performance
Security

24. 3.
Node.js
Functional Programming

25. Functional Programming
✘ Composition
Node.js

26. Functional Programming
✘ Composition
Asynchronous control flow
✘ Generators
Node.js

27. Node.js
const brequest = yieldable(throttle(retry(breaker(batch(request)))));
const response = yield brequest.get(...);

28. 4.
Database
Partitioning & operations

29. Data Size
✘ Small documents
✘ TB / day
Datasets

30. Partitioning

31. Benefits
✘ Manageability
✘ Performance
✘ Availability
✘ Load balancing
https://en.wikipedia.org/wiki/Partition_(database)
Partitioning

32. Keys
✘ App-level
✘ Time-based
Partitioning

33. Keys
✘ App-level
✘ Time-based
Retention
✘ Backup / restore
✘ Delete
Partitioning

34. 5.
Monitoring
health & metrics

35. Health
Traffic
Circuit state
Error percentage
Hystrix Streams

36. Hystrix
Dashboard

37. Hystrix + Turbine + Eureka
Turbine
Aggregate
data
Hystrix
Data
streams
Eureka
Service
discovery

38. Direct communication
✘ Networking release
Eureka client
✘ Code instrumentation
Service Discovery

39. Mozzle
https://github.com/Bo0mer/mozzle

40. Mozzle + RIG
Riemann +
InfluxDB
Store data
Mozzle
Pull metrics
from CF
Grafana
Visualize

41. CF metrics
in
Grafana

42. 6.
Alerts
on problems

43. Monitoring & Alerts
Alerts

44. Monitoring & Alerts
✘ Reactive
Alerts

45. Monitoring & Alerts
✘ Reactive
Anomaly Detection
✘ Pro-active
Alerts

46. 7.
Deployment
to Cloud Foundry

47. Applications vs. Processes
12-factor apps
✘ Factor VI: Processes
Abacus: set of processes
CF: single-process apps

48. Applications vs. Processes
Process support
✘ deployment
✘ management
✘ flexibility

49. Distributed applications are hard
✘ network
✘ monitoring
Take-aways

50. Distributed applications are hard
Node.js: developing cloud-native apps
✘ async
✘ composition
✘ generators
Take-aways

51. Distributed applications are hard
Node.js: developing cloud-native apps
Cloud Foundry: running cloud-native apps
✘ container networking
✘ deployment & operations
Take-aways

52. thanks!
Any questions?
You can find us at
hristo.iliev@sap.com, hsiliev@gmail.com
stoyan.rachev@sap.com, stoyanr@gmail.com

53. Credits
Special thanks to all the people who made and released
these awesome resources for free:
✘ Presentation template by SlidesCarnival
✘ Photographs by Unsplash

54. SlidesCarnival icons are editable shapes.
This means that you can:
● Resize them without losing quality.
● Change fill color and opacity.
Isn’t that nice? :)
Examples:

55. Now you can use any emoji as an icon!
And of course it resizes without losing quality and you can change the color.
How? Follow Google instructions
https://twitter.com/googledocs/status/730087240156643328
✋ ❤
and many more...

56. Extra graphics