Presentation is loading. Please wait.

Presentation is loading. Please wait.

Usage Compute Time Average Inactivity Period Compute Time Average Usage Compute Time Compute Time Average Usage.

Similar presentations


Presentation on theme: "Usage Compute Time Average Inactivity Period Compute Time Average Usage Compute Time Compute Time Average Usage."— Presentation transcript:

1

2 Usage Compute Time Average Inactivity Period Compute Time Average Usage Compute Time Compute Time Average Usage

3 BING SEARCHES – JAPAN VS. UK Source: Microsoft

4 turbotax.com taxcut.com hrblock.com taxact.com Source: Alexa ~4x normal load (Holiday shopping) ~10x normal load (Tax season) target.com walmart.com toysrus.com barnesandnoble.com Source: Alexa

5

6 Business logic Datacenter (Power and Cooling) Respond to hardware failures Monitoring and alerting infrastructure Reliable/Secure storage and computation Metering and billing infrastructure Live upgrades and OS patches Add compute/storage capacity on the fly Overprovision for peak traffic Service “glue” … Buy and provision hardware

7 …. Service 1Service 2Service NService 3 ……

8

9 2) Choose image, then create and configure VM(s) for application 1) Choose image, then create VM for DBMS and configure DBMS Library VM Images Developer/Ops Application Data Load Balancer 5) Configur e load balancer 6) Manage VMs and DBMS (e.g., deploying new OS images in VMs) 3) Provision database, then create tables and add data 4) Install applicatio n

10 Developer/O ps Application Data Load Balancer 2) Deploy application w/ service model 1) Provision database, then create tables and add data 3) Automated Service Managemen t

11

12 Windows Azure Networking

13

14 Highly-available Fabric Controller Hardware control Software control WS08 Hypervisor VM Fabric Agent Switches Load- balancers

15

16

17

18 Network Load Balancer Stateless ‘Worker’ Stateless Front End Shared Filesystem (Azure Blobs) Partitioned RDBMS (SQL Azure) Key/Value Datastore (Azure Tables) Azure Queues

19 Durabl e Store

20 Front- End Window s Azure Storage, SQL Azure Window s Azure Storage, SQL Azure Load Balancer Middle- Tier

21 Service Model Role: Front-End Definition Type: Web VM Size: Small Endpoints: External-1 Configuration Instances: 2 Update Domains: 3 Fault Domains: 2 Role: Middle-Tier Definition Type: Worker VM Size: Large Endpoints: Internal-1 Configuration Instances: 3 Update Domains: 3 Fault Domains: 2

22

23

24

25 Allocation across fault and update domains Load-balancers

26

27

28 Availability %Downtime per yearDowntime per month*Downtime per week 99% ("two nines")3.65 days7.20 hours1.68 hours 99.9% ("three nines")8.76 hours43.2 minutes10.1 minutes 99.99% ("four nines")52.56 minutes4.32 minutes1.01 minutes 99.999% ("five nines")5.26 minutes25.9 seconds6.05 seconds

29

30 TOR LB Agg PDU LB Agg LB Agg LB Agg LB Agg LB Agg Racks Datacenter Routers Aggregation Routers and Load Balancers TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU TOR PDU ………… … Top of Rack Switches Power Distribution Units … Nodes Top of Rack Switch is a Single Point of Failure

31 Front- End-1 Front- End-2 Middle Tier-2 Middle Tier-1 Middle Tier-3 Front- End-1 Middle Tier-1 Front- End-2 Middle Tier-2 Middle Tier-3

32 Upgrade domains allocated across fault domains Fault domains

33

34

35

36

37

38 Role B Middle-Tier Role Count: 3 Update Domains: 3 Size: Large Load Balancer Fault domain Upgrade domain

39 Fabric Controller (Primary) FC Host Agent Host Partition Guest Partition Guest Agent Guest Partition Guest Agent Guest Partition Guest Agent Guest Partition Guest Agent Physical Node Fabric Controller (Replica) … Role Instance Trust boundary Image Repository (OS VHDs, role ZIP files)

40

41 ProblemFabric DetectionFabric Response Role instance crashesFC guest agent monitors role terminationFC restarts role Guest VM or agent crashesFC host agent notices missing guest agent heartbeats FC restarts VM and hosted role Host OS or agent crashesFC notices missing host agent heartbeatTries to recover node FC reallocates roles to other nodes Detected node hardware issueHost agent informs FCFC migrates roles to other nodes Marks node “out for repair”

42

43 Front- End-1 Front- End-2 Update Domain 1 Update Domain 2 Middle Tier-1 Middle Tier-2 Middle Tier-3 Update Domain 3 Middle Tier-3 Front- End-2 Front- End-1 Middle Tier-2 Middle Tier-1

44

45


Download ppt "Usage Compute Time Average Inactivity Period Compute Time Average Usage Compute Time Compute Time Average Usage."

Similar presentations


Ads by Google