Avi Deployment Guide for Microsoft Azure
About Microsoft Azure
Microsoft Azure is a cloud computing service that offers hosting and related public cloud services, as well as developer products to build a range of programs from simple websites to complex applications.
Azure provides a set of modular cloud-based services with a host of development tools, including hosting and computing, cloud storage, data storage, translation APIs and prediction APIs. Figure 1 depicts a sample Azure deployment.
Figure 1. Azure deployment
About Avi Vantage
The Avi Vantage Platform provides enterprise-grade distributed ADC solutions for on-premises as well as public-cloud infrastructure. Avi Vantage also provides built-in analytics to diagnose and improve the end-user application experience, while making operationalizing easier for network administrators.
Avi Vantage is a complete software solution which runs on commodity x86 servers or as a virtual machine and is entirely enabled by its REST API.
Purpose of This Guide
This document describes the process of provisioning and configuring Avi Vantage as an application delivery controller for application workloads running inside Azure.
The document is intended for
- Network administrators: To configure and operationalize the Avi Vantage solution.
- Azure system administrators: To provision the Avi Vantage solution.
We assume familiarity with
- The basics of load balancing and application delivery.
- Basic Azure functionality. For detailed information refer to Microsoft Azure Documentation.
Use of Avi Vantage with Azure provides the following functionality:
- The Avi Vantage Controller is available as an Azure appliance (VHD).
- Once the Avi Controller is deployed, Azure account details and credentials are provided to it. It then connects to the Azure infrastructure and automatically provisions Service Engines as required.
- A single interface is available on the SE for control and data traffic (in-band management).
- VIP addresses are allocated from Azure IPAM.
- An optional, public VIP can be allocated automatically to a virtual service, along with a private VIP address.
Prerequisites and Assumptions
Both Microsoft Azure and Avi Vantage provide a variety of configuration and deployment options, based on individual requirements. This guide makes the following assumptions regarding the infrastructure:
- For resource group where the Avi Controller is spawned, a role of contributor or higher is required.
- For virtual network where the Avi Service Engine instances are to be deployed, a role of AviController or higher is required. For more details on creating the AviController role, refer to Role Setup for Installation into Microsoft Azure.
- Specific ports need to be allowed on the Service Engine and Controller management subnets to enable Controller-to-Service Engine communication. For details, refer to Protocol Ports Used by Avi Vantage for Management Communication.
- The Service Engine subnet should allow incoming TCP connections on port 7 from the IP address 220.127.116.11. This is used by Azure to probe the Service Engine health. For more details on this requirement, refer to Understand load balancer probes.
The resource group must have an Azure Virtual Network (VNet) configured with a subnet.
For the purpose of this document, the resource group avi-vantage will be used to deploy the Avi solution. As displayed in the screenshot below, this group has
avi-vantage-vnet VNet, with an available address space of
10.20.0.0/16 and a subnet of
Avi Controller Instantiation
Avi Vantage is available in Azure Marketplace as a Bring Your Own License (BYOL) offering.
Marketplace Link : Navigate to the Avi Vantage page on Azure Marketplace.
Click on Get it Now to start the deployment process. If the deployment is via the Azure portal then create a new VM and search for Avi Networks. The Avi Vantage VM will show up in the search results.
Follow the given steps to initiate the deployment:
Provide the information requested under the Basics tab.
Click on OK to continue to the next tab.
- Based on deployment scale considerations, choose an appropriate VM size. The following table lists the minimum requirements for the VMs on which the Avi Controller and Avi SEs are installed.
Component Memory vCPUs Disk Avi Controller 24 GB 8 64 GB Service Engine 2 GB 2 10 GB
For Avi Controller, we recommend the following instance types:
Deployment Size Virtual Service Scale Instance Type Memory vCPUs Disk (Minimum) Small 100 F8S 16 GB 8 64 GB Medium 1000 F8S 16 GB 8 64 GB Large 5000 F16S 32 GB 16 64 GB
Refer to the Disk Capacity Allocation section in the Avi Controller Sizing KB for recommended hard disk size. The below example shows a choice of eight CPUs and 16 GB memory. (Instance: F8S)
In the Settings tab, select the following options:
- Availability set: It is recommended to use an availability set for Avi Controllers.
- Storage : Select Yes for managed disks.
- Virtual network: Create a new VNet, or use an existing VNet.
- Subnet: Select a subnet for Avi controller management IP address to be allocated from.
- (Optional) Public IP address: Allocate an existing or new public IP address to the controller VM.
- Network security group (firewall): Apply an existing or new network security group to restrict traffic to the controller.
Click on OK followed by Purchase to run final validations and initiate the deployment.
(Optional) Create a Controller cluster
To ensure complete redundancy, two additional Avi Controller nodes can be added to create a 3-node Avi Controller cluster.
To create a Controller cluster, use the Azure template available at the link here.
Once the Controller is up, it can be configured via a web browser. The FQDN will be mentioned as an output of the template execution, as in this case, avicontrollerpubip.westus.cloudapp.azure.com.
Avi Vantage Configuration
Follow the given steps to complete the initial configuration. Each step is provided with an associated screenshot.
Provide credentials for the administrator account (Username:
Provide DNS and NTP Settings (Can be edited later).
Provide an email address to be used for alerts from the controller (Can be set up at a later stage).
Select No Orchestrator to complete the initial configuration.
Continue by clicking on No for Support multiple Tenants (Multi tenancy can be enabled later).
Once the setup is completed, the browser will automatically refresh to the Avi Controller dashboard.
Configuring Azure Cloud
At this point, the Avi Controller is provisioned but not connected to any ecosystem. The next step is to create a cloud configuration of type Azure, so that Avi Vantage can spin up Service Engines in the Azure VNet, and the load balance workloads present there.
Follow the given steps to complete the cloud configuration. Each step is provided with an associated screenshot.
Click on the Applications tab and navigate to Infrastructure -> Clouds
Click on the Create button to add a new cloud. Provide a name, and select Microsoft Azure as the Cloud infrastructure type.
On the next tab, provide information related to the Azure account.
Start by clicking on Create Credentials tab and provide Azure credentials.
You can either choose an Azure account username/password, or an Application ID. In the screenshot below, the username method is used.
Save and select these newly created credentials and provide the Azure subscription ID. Click Next.
Provide the Azure location details. These details are associated with the location of resource group, the resource group and VNet that can be used, and the subnet for Service Engine management network.
Optionally, a DNS provider can be selected as well. Instead of Azure DNS, AWS Route 53 can also be used by selecting Other.
Click on Complete, to provision the Azure cloud. At this time, the Controller will upload the Service Engine VHD into an Azure storage account, so that SEs can be deployed as required by the applications.
Save the settings. The system is now ready for virtual service creation.
Virtual Service Configuration
To create a virtual service to load balance an application workload, perform the following steps:
- Create a pool containing application servers that need to be load balanced.
- Create a virtual service with a front-end virtual IP.
Navigate to Application -> Pools and click Create Pool.
Provide a pool name. The other fields are optional and the defaults are sufficient. Click Next.
Add one or more application (back-end) servers. If the applications are part of an Azure scale set, the scale set option can be selected. If not, just provide the IP addresses of the servers and click Next.
Click through the remaining steps, by retaining the defaults to complete the pool creation process.
Creating the Virtual Service
Navigate to Application -> Virtual Services and click Create Virtual Service. Select Advanced Setup.
Provide a VS name.
Select a network from which the front-end VIP should be allocated. The VIP will be allocated by Azure.
If the virtual service needs to be accessible via the Internet, select the option Assign Public IP for External Client Access.
Select the service ports. Port 80 is configured by default. Add port 443 as an SSL port as well.
In the Pool section, select the previously created pool from the dropdown menu.
Click Next through the remaining screens. Click Save at the last screen to complete the provisioning.
At this point, the UI will refresh to the VS dashboard.
The Avi deployment and virtual service configuration is now complete. Wait for 2-3 minutes for the internal Azure network configurations to be completed, before sending traffic for verification. Send some traffic from a client to the virtual service IP to verify if the virtual service is functioning.
Azure VM Sizes for the Avi Service Engines
Avi Service Engines are automatically deployed on Azure by the Avi Controller, based on the virtual services that have been configured.
Avi SEs can be deployed on VMs with various sizes. This can be configured under Service Engine Group -> Advanced setting.
The table below shows the maximum SSL TPS performance observed on some Azure VM sizes.
|Azure VM Size||SSL TPS Performance|
- The performance results provided above are indicative numbers for a subset of instance types. There are other VM sizes available under the Service Engine group settings that can be used.
- SSL performance (TPS - transactions per second) has been measured considering one configured virtual service (HTTPS, ECDHE-ECDSA-AES256-SHA cipher) and GET requests for a 128-byte payload without session reuse. More details regarding Service Engine performance can be found here.
Updated: 2017-12-18 09:37:42 +0000