Presented by Andrey Nikiforov
This talk covers the topics around cloud planning and architectures. Covering how to achieve attaining the various abilities, understand the alternatives, popular frameworks, etc.
2. Agenda
• Defining the Cloud
• Developer Point of View: coding for Azure
• Designing for the Cloud
• Value Propositions
3. Cloud and *aaS
An approach to computing that enables applications
to be delivered at scale for a variety of workloads and client devices
Office 365, CRM, Any Web App You Are
SaaS
Building
PaaS Web, Storage, ESB, Authentication
IaaS CPU, Network, HDD
Flexibility Offerings
4. Cloud Services: Roles vs. Application Types
PaaS IaaS
Web Role Worker Role VM Role Cloud
Web Windows
Application Service On Premises
5. Services vs. Apps
Queue
Service
SQL Bus Cache NoSQL AuthN CDN VPN
SQL Queue Cache Table ACS CDN Connect Cloud
Service BLOB
Bus
SQL
SQL MSMQ Velocity AD ? On Premises
6. SQL Database
• High Availability (triple replicated)
• No Physical Management
• No Agent, Notification broker, etc,
• No Context Switching (no USE)
• No Distributed Transactions
• SQL AUTHN only
TDS
TDS Load Balancer
7. Storage Services
• BLOB, Table, Queue
• Triple replicated
• Geo replicated (BLOB & Table) asynchronously
• In PROD since 2008
• Used by MS internally (e.g. Bing)
• BLOB -> Azure Drive
8. Storage Services: BLOB
• Block (200G)
– Can be locked exclusively
– Blocks have MD5 and up to 4M
– Transactional (upload blocks before committing)
– Optimized for streaming
• Page (1T)
– In-place updates
– Page is 512 byte
– Can write group of pages
– Optimized for random access
• Snapshots
• Controlled Shared Access
• CDN
9. Storage Services: Table
• Key-Value Pair Storage on Steroids
– Key==PatritionKey+RowKey
– Value==Entity with up to 255 properties and up to 1M
• Up to 100T
• Supports Continuation for Queries
• Entity Group Transactions
– Same Partition
– <100 Entities
– <4M in Size Total
10. Storage Services: Queue
• Durable Storage of Messages, up to 8K in size each
• FIFO is not guaranteed
• “At Least Once” delivery
11. Azure App from Developer Point of View
• Same Web App
• Worker Role is very similar to Win Services
• Services available through
– REST API
– .NET REST wrappers
• Diagnostics
– Usage is the same (E.g. Trace.TraceInformation())
– Data Collection is different
• Configuration management enhancements
12. Developer Tools
• Azure Tools for Visual Studio
– Project Templates
– Compute Emulator
– Storage Emulator (Table, Blob, Queue)
• SQL Server
13. Demo
• Create an Azure app
• Try it in Emulator
• Deploy to Azure
– Environment configurations
– Deployment profiles
• Add Instrumentation
– DiagnosticsMonitor
15. Value Propositions
• Assets -> Operation Expenses
• No Overprovisioning
• Expenses Are Easily Tightened to Sales/Activity
• Simplified Dev Process Support (Dev/QA provisioning)
16. Action Plan
• Get Azure Trial Subscription
• Get Azure Labs
• Write your own BeyondHelloWorld
– Different services (e.g. SQL & Tables)
– Measure performance/scalability
– Calculate cost
17. Summary
• What is it “Cloud”
• Azure Services
• Easy Development for Azure
• Value Propositions
Notes de l'éditeur
Cloud: simply an approach to computing that enables applications to be delivered at scale for a variety of workloads and client devicesScale on the slide: == “how much consumer manages”. Saas == very little by consumer and a lot for consumer by service provider