posted by George_Curelet_Balan

27:20 Cloud challenge nr 1: Scale
  lots of users, data, always busy: how to think different about scalability
    Scale up is not the solution. Scale out is.
29:06 a. Scale the services: First scale out option:
        partition the app in:
          as set of async loosely coupled services
     31:15 example of message coupled sell & inventory services

35:20 The message based design pattern allows possible release of part of the software. Cloud changes the way we think about software.

36:20 b. scale data. SQL Azure Federations supports "sharding" cloud scale data

partioned across multiple physical tables
  partition diff tables by region
    very hard problem

39:30 c. State maintanance in scale scenario
 big challenge to keep cloud elastic + solutions (minimize & isolate state in the service)
 AppFabric cache services to the rescue

42:00 Performance becomes explicit in the cloud
  because of charging model for cloud usage (storage, round-trips, etc.)

46:40 cost of operation affects sys design
   cost of SQL Azure storage is higher than windows blob storage

47:10 service success is determined by its manageability
 "One Microsoft" cloud experience with different MS enterprise products

50:20 provide service visibility to the customer (up-time, scheduled events, etc.)

51:10 Cloud error reporting and propagation adds new complexity
 need for platform services & APIs for error reporting

52:40 cost effective manageability of service

53:50 new reality: combination of cloud &
                private devices (on-premises due to regulatory, merging, etc.)

55:30 Identity

56:10 Equivalent of liveID in the enterprise envir (online ID, bussiness equivalent available in the cloud, similar to AD)

57:50 connect scenarios from cloud to premises
   azure service bus, CRM eventing model

58:25 Customization for Cloud Systems
 declarative model-driven customization

1:02:53 Other Cloud Considerations
   focus on simplicity
       should be rich but easy to startup, customize & upgrade
       remove features

   Privacy & data security
     raised to new levels

   Component failure
      assume failure due to huge nr of nodes
      recover state info in case of failure

1:09:30 on-premises customers benefit from cloud investment

1:11:23 The cloud enables new scenarios
 shared data
   payroll regulation
 connections
  connected big supply chains
    limit number of interfaces if everyone talks with a cloud hub
 new analytics, reveals otherwise invisible trends
  across massive data in the cloud
    example: determine flu trends via queries
  
1:18:10 Q&A

1:22:38 how MS deals with competitors?

posted by George_Curelet_Balan