Tuesday, May 15, 2018

NSX integration with vRealize Automation 7.4 - part 2

In part 1 of this post we presented the configuration at vRA level. In this post we'll see how to create a service in the Service Catalog for  programmatic NSX consumption.

First let's remember the main concepts of vRealize Automation service catalog:

  • catalog items are published in the service catalog for user consumption e.g Linux VM or 3 tier web app
  • catalog items can be grouped under different services: QA, Test&Dev, Web Apps, Linux Servers
  • a user is allowed to request a service item based on his entitlements; Entitlements define who has access to catalog items and what actions he can do 

For start, we'll create a service called Linux VMs and a new Entitlement called Allow Linux VMs. We'll entitle all users of the business group to the Linux VMs service. Using services in the entitlement instead of individual items we make sure that every new item mapped to this service will be automatically accessible to the users.

Administration > Catalog Management > Services

Administration > Catalog Management > Entitlements

Next we'll create a blueprint that deploys vSphere VMs. There are several ways to provision vSphere VMs, we will use linked clones because they are very fast and use deltas to keep the changes (which is good in labs). To use linked clones we need to create a golden image: a VM configured to the desired state.

First create the VM: deploy it from an existing template or create it from scratch. VM hostname and networking details will be configured at deployment during guest OS customization. For this to work we need VMware tools installed in the VM and a customization specification created in vCenter Server. 

No other special configuration is needed for the VM.

Optional step (vRA agent installation): if you don't plan to run scripts inside the guest OS of the vRA manged VM, you can skip this step. The installation should be pretty easy since VMware already provides a script that can handle it. Go to vRA appliance URL and download the script on your Linux VM:

 wget  https://vra_app_fqdn/software/download/prepare_vra_template_linux.tar.gz --no-check-certificate

Then extract the script from the archive and run it:

tar -xvzf prepare_vra_template_linux.tar.gz
cd prepare_vra_template_linux

Choose default agent type (vSphere), add the address of vRealize Appliance, manager service, accept the key fingerprints for the certificates, set the download timeout, and install JRE (if not already in the VM)

Now we have a VM with all the utils inside (VMware tools and optionally vRA agent) and we create the snapshot that will be the base for linked clones.

At this point we login to vRA portal and we start working on our service creation. Go to Design > Blueprints. Start creating a New blueprint. Type the name of the blueprint, assign an unique ID or leave the automatically generated one, limit the number of deployments per request (if you want). Add lease days to control the spawn of deployed VMs (especially for temporary environments) and add a period of time you want the item to be archived before deleting (when lease expires).

Since this is for demo, I've added a default lease of 1 day and no archival (automatic deletion after lease expires). On NSX Settings tab, choose the NSX transport zone and if you want to isolate the VMs deployed from this blueprint (allow only internal traffic between VMs).

Pressing OK button will take you to the canvas. From Machine Types category, drag and drop vSphere (vCenter) Machine

From Network&Security category, drag and drop On-Demand Routed Network

Select the vSphere__vCenter__Machine_1 components form the canvas and fill in the configuration details. Add the number of instances that can be deployed in a request 

Add build information: how the VM will be created (linked clone), where to clone it from, what customization specification to use:

Type the VM consumed resources: number of CPUs, memory, storage. Take care when configuring these values: if you allow 10 instances in a deployment, each instance with a maximum of 8 vCPU and 32 GB of RAM you may end up with a deployment using 80 vCPU and 320 GB of RAM. This is a good moment when approval workflows come into place.

Finally we need to connect the VM to the network. But first we'll configure the network component. On canvas select the On-Demand_Routed_Network_1 components and choose the parent network profile (profile that has been created in part 1

Go back to vSphere component, go to Network tab and click New. From drop down box select the network name

Lastly, add a custom property for the VM to define the operating system that is being used

At this moment we've configured how to create the VM, how to create the network and linked the VM to the network. Press Finish and then Publish the blueprint:

Once the blueprint has been published, it will appear under Administration > Catalog Management > Catalog Items. Select the new catalog item, press Configure and map it to the service created earlier at the beginning of the post. 

The service will appear in the Catalog tab and you can press Request to deploy a new instance of it. To see what is happening, go to Requests tab, select the request, press View Details and when the request details open press Execution Information

Here you will see that the vxlan has been created on demand and DLR reconfigured. Also the VM has been created and mapped to the new vxlan. The process can also be monitored in vCenter Server

After the provisioning finished successfully, the components will be displayed in Items tab from where they can be managed using day 2 operations.

No comments: