Didn’t make it to TechEd this year? Don’t worry! This month, we are releasing a new article that highlights the Best of TechEd every workday. In today’s article I look at Mark Russinovich’s talk on Azure Internals. Watch his session and learn about Microsoft’s data center architecture, what goes on behind the scenes when you deploy and update a Windows Azure app, and how it monitors and responds to the health of machines, its own components and the apps it hosts. Mark also highlights recent enhancements and how the platform implements some of the latest features.
Here are a few things that I found interesting in this session:
- The Fabric Controller- Manages everything in a Cluster (about 1,000 servers)
- The Datacenter.xml file has the description for the 1,000 machines, their power strips, and their networks.
Steps when Provisioning a Physical Node
- Physical Machine Powers On and does a PXE-Boot
- A WinPE instance loads – which downloads a VHD file to the local attached storage
- Physical Machine reboots and boots from the VHD file (can cache multiple vhd boot images)
- This server can be multiple things. One of the things it could be is a Fabric Controller server, it could also be a Hyper-v server.
Hardware load balancers were expensive and not able to handle the needs of the Azure datacenter so they developed a Software Load Balancer and that feature is being added to Server 2012R2.
Here is the link to the Windows Azure Internals session by Mark Russinovich:
Here are a few interesting spots in the session
35:30 – How Azure provisions a node (server)
39:10 – Deploying a Service
48:40 – Demo – view of a service across different fault domains in a cluster
51:02 – Visualize a service deploying across 3 servers
52:19 – Deploying a Role Instance – Differencing Disks created
55:20 – Infrastructure as a Service – Topology
1:03:32 – Maintaining Service Health
1:08:10 – Azure Operations Room Video ( 2 guys manage all of Azure!)