Building scalable cloud-native applications (Sam Vanhoutte at Codit Azure PaaS Event)
Download as PPTX, PDF0 likes359 views
This document discusses learnings from building scalable cloud-native solutions. It covers considerations for scalability like decoupling services, partitioning data and throttling external communications. Specific patterns for communication between services are examined like asynchronous messaging for durability and load leveling. Testing is emphasized to ensure solutions can scale out. Designing for changes in the cloud over time with new services and features is also advised.
Building scalable cloud-native applications (Sam Vanhoutte at Codit Azure PaaS Event)
- 1. Building scalable, cloud-native solutions. Learnings from the field. SamVanhoutte, CTO @samvanhoutte
- 2. Hello world 2 3rd year Azure MVP + 100Active customers 2000 Belgium 2004 France 2008 Netherlands 2013 Portugal 2016 Switzerland 2016 U.K. Close collaboration with Microsoft +120 employees Azure & IoT Advisor 8th year Microsoft P-TSP Azure CAT support Integration Azure IoT API mgmt sam.vanhoutte@codit.eu +32 474 849 993 @SamVanhoutte be.linkedin.com/in/samvanhoutte
- 3. Agenda
- 4. Why is integration more important now than ever? Architecture is key, integration is the backbone
- 6. The interface is where the profit is the world’s most popular media company creates no content the most valuable retailer carries no stock the world’s largest accommodation provider owns no property the world largest taxi company owns no car
- 7. The story of the WiFi-Kettle 7 A British guys wanted his morning tea.
- 8. The story of the WiFi-Kettle 8 Luckily he was a geek and decided to live tweet his 11hr struggle.
- 9. The story of the WiFi-Kettle 9
- 10. The story of the WiFi-Kettle 10
- 11. It all starts with an idea 11
- 12. Learnings from the field 12
- 13. Consumption prediction in Azure PaaS Prediction (start) vs Reality (go-live)
- 14. Considerations that impact scalability decouple, partition & throttle Throttling External communications Partitioning MicroservicesResource size Concurrency & locking Communication patterns Caching
- 15. Chosing the right communication pattern 15 Information lookup MostWeb API’s Avoid data duplication UI integration RPC Async Durable messaging Scalable & multi-cast Load leveling Temporary unavailable Bulk Scheduled data exchange Bulk or batch data Traditional apps Initial import & data load
- 16. Handle external communications 16 Requests outside of app Security, monitoring Documentation, portal DNS handling API Management Queuing Standards based Durable messaging Load leveling Peek/lock/retry Version API versioning Message versioning Monitor usage
- 17. ➔ Testing is crucial ➔ Mostly scaling out is easier than scaling up ➔ Autoscaling is welcome, but complex Resource sizes matter 17 Many small vs Few large The rise of serverless Autoscale & test
- 18. ➔ Think about downstream (LOB?) systems ➔ Aggregate and forward, when possible ➔ Retries are important, but think about your strategy Throttling will help your connected systems 18 Async messaging Plan retries Timers & Reminders
- 19. ➔ Synchronization of cache & data store ➔ Think distributed Cache, but not too much 19 Distributed cache Write through Read behind Stateful actors or services
- 20. ➔ Use hashing algorithm, based on functional value ➔ Stateless vs stateful Data & service partioning 20 Colocated data Parallellism over partitions Partitioning scheme
- 21. Multi tenancy Isolate data, but allow for anonymized cross tenant analytics
- 22. Track & trace 22 Tracing frameworks ➔ Opt for centralized monitoring ➔ Extract ‘functional’ metadata Correlation Management Application insights
- 23. DevOps 23 More self-service Security & RBAC Staging & ALM External connections ➔ Align procedures ➔ Good communication between Dev & Ops ➔ Automation of ALM
- 24. Fast iterations in the cloud 24 New services New features Backwards compatibility Deprecation of services ➔ Closely follow up on roadmap & announcements ➔ Design for change ➔ Use abstraction to cope with changes ➔ Foresee versioning or compatibility for externalAPI
Editor's Notes
- #4: Multi tenancy Tracing & logging API management from the beginning Scalability and partitioning Security concerns Pricing Stateful vs stateless Serverless Scalability (small vs large) Geo-deployments (traffic manager / cdn) Concurrency & locking when scaling Throttling
- #5: Businesses are seen as being innovative IT is seen as catching up to innovation 3 components: proliferation of apps, data as management nightmare, smart devices that need to interact
- #6: Forrester report that surveyed 675 cloud developers 39% delivering an application integration solution Higher than mobile, e-commerce
- #7: control the interface between the consumer and the provider take a cut from the millions of consumers
- #14: Businesses are seen as being innovative IT is seen as catching up to innovation 3 components: proliferation of apps, data as management nightmare, smart devices that need to interact
- #15: X API management from the beginning (external buffer) X Scalability and partitioning X Microservices - Stateful vs stateless Serverless (attention points) X Scalability (small vs large) X Concurrency & locking when scaling X Throttling
- #22: Businesses are seen as being innovative IT is seen as catching up to innovation 3 components: proliferation of apps, data as management nightmare, smart devices that need to interact
- #23: CIC demo Refer to ALM
- #24: RBAC in Azure portal
- #25: API management