Merge pull request #77645 from SyntaxC4-MSFT/live

Gartner Hotfix

Author: SyntaxC4

articles/networking/azure-network.md

@@ -22,14 +22,18 @@ Every day, customers around the world connect to Microsoft Azure, Bing, Dynamics
 
 The Microsoft global wide-area network (WAN) plays an important part in delivering a great cloud service experience. Connecting hundreds of datacenters in 54 regions around the world, our global network offers the availability, high capacity, and the flexibility to respond to unpredictable demand spikes. 
 
-## Best network experience out of the box 
+## Best network experience out of the box
+ 
 You want a fast, reliable response when you use Microsoft services. Data travels over our network at nearly the speed of light; network speed, or latency, is a function of distance from the customer to the datacenter. 
 
 Customer traffic enters our global network through strategically placed Microsoft Edge nodes, our points of presence. These edge nodes are directly interconnected to more than 3500 unique Internet partners through thousands of connections in more than 145 locations. Our rich interconnection strategy optimizes the paths that data travels on our global network. Customers get a better network experience with less latency, jitter, and packet loss with more throughput. Direct interconnections give customers better quality of service compared to transit links, because there are fewer hops, fewer parties, and better networking paths. 
 
-Say you’re in London and the service is in Tokyo: Internet traffic enters one of our edges in London, goes over Microsoft WAN through France, our Trans-Arabia paths between Europe and India and then to Japan where the service is hosted. Response traffic is symmetric. This is sometimes referred to as [cold-potato routing, which means that traffic stays on Microsoft network as long as possible before we hand it off. 
+Say you’re in London and the service is in Tokyo: Internet traffic enters one of our edges in London, goes over Microsoft WAN through France, our Trans-Arabia paths between Europe and India and then to Japan where the service is hosted. Response traffic is symmetric. This is sometimes referred to as [cold-potato routing](https://en.wikipedia.org/wiki/Hot-potato_and_cold-potato_routing), which means that traffic stays on Microsoft network as long as possible before we hand it off. 
 
-Azure traffic between our datacenters stays on our network and does not flow over the Internet. This includes all traffic between Microsoft services anywhere in the world. For example, within Azure, traffic between virtual machines, storage, and SQL communication traverses only the Microsoft network, regardless of the source and destination region. [Intra-region VNet-to-VNet traffic](../virtual-network/virtual-network-peering-overview.md), as well as [cross-region VNet-to-VNet traffic](../vpn-gateway/vpn-gateway-howto-vnet-vnet-resource-manager-portal.md), stays on our secure Microsoft network. 
+Azure traffic between our datacenters stays on our network and does not flow over the Internet. This includes all traffic between Microsoft services anywhere in the world. For example, within Azure, traffic between virtual machines, storage, and SQL communication traverses only the Microsoft network, regardless of the source and destination region. [Intra-region VNet-to-VNet traffic](../virtual-network/virtual-network-peering-overview.md), as well as [cross-region VNet-to-VNet traffic](../vpn-gateway/vpn-gateway-howto-vnet-vnet-resource-manager-portal.md), stays on our secure Microsoft network.
+
+![Microsoft global WAN](./media/azure-network/microsoft-global-wan.png)
+*Figure 1: Microsoft global WAN*
 
 To support the tremendous growth of our cloud services and maintain consistent service level agreements, we have invested massively in fiber capacity and diversity in metro, terrestrial and submarine paths. Our [MAREA](https://www.submarinecablemap.com/#/submarine-cable/marea) submarine cable, industry’s first Open Line System (OLS) over subsea, between Bilbao, Spain and Virginia Beach, Virginia, USA is live as well as [AEC](https://www.submarinecablemap.com/#/submarine-cable/aeconnect-1) between New York, USA and Dublin, Ireland and [New Cross Pacific (NCP)](https://www.submarinecablemap.com/#/submarine-cable/new-cross-pacific-ncp-cable-system) between Tokyo, Japan, and Portland, Oregon, USA.
 

articles/networking/media/azure-network/microsoft-global-wan.png

@@ -96,6 +96,22 @@ A snapshot is a copy of a disk at the point in time the snapshot is taken. It ap
 
 A snapshot doesn't have awareness of any disk except the one it contains. This makes it problematic to use in scenarios that require the coordination of multiple disks, such as striping. Snapshots would need to be able to coordinate with each other and this is currently not supported.
 
+## Disk allocation and performance
+
+The following diagram depicts real-time allocation of bandwidth and IOPS for disks, using a three-level provisioning system:
+
+![Three level provisioning system showing bandwidth and IOPS allocation](media/virtual-machines-managed-disks-overview/real-time-disk-allocation.png)
+
+The first level provisioning sets the per-disk IOPS and bandwidth assignment.  At the second level, compute server host implements SSD provisioning, applying it only to data that is stored on the server’s SSD, which includes disks with caching (ReadWrite and ReadOnly) as well as local and temp disks. Finally, VM network provisioning takes place at the third level for any I/O that the compute host sends to Azure Storage's backend. With this scheme, the performance of a VM depends on a variety of factors, from how the VM uses the local SSD, to the number of disks attached, as well as the performance and caching type of the disks it has attached.
+
+As an example of these limitations, a Standard_DS1v1 VM is prevented from achieving the 5,000 IOPS potential of a P30 disk, whether it is cached or not, because of limits at the SSD and network levels:
+
+![Standard_DS1v1 example allocation](media/virtual-machines-managed-disks-overview/example-vm-allocation.png)
+
+Azure uses prioritized network channel for disk traffic, which gets the precedence over other low priority of network traffic. This helps disks maintain their expected performance in case of network contentions. Similarly, Azure Storage handles resource contentions and other issues in the background with automatic load balancing. Azure Storage allocates required resources when you create a disk, and applies proactive and reactive balancing of resources to handle the traffic level. This further ensures disks can sustain their expected IOPS and throughput targets. You can use the VM-level and Disk-level metrics to track the performance and setup alerts as needed.
+
+Refer to our [design for high performance](../articles/virtual-machines/windows/premium-storage-performance.md) article, to learn the best practices for optimizing VM + Disk configurations so that you can achieve your desired performance
+
 ## Next steps
 
-Learn more about the individual disk types Azure offers and which type is a good fit for your needs in our article on disk types.
+Learn more about the individual disk types Azure offers, which type is a good fit for your needs, and learn about their performance targets in our article on disk types.