vRealize Automation 8 First Impressions – Cloud Assembly

The Cloud Assembly section of vRealize Automation 8 is the one that vRA Administrators will most likely spend the most time. In vRA 7 terms, it constitutes aspects of the Infrastructure and Adminstration areas, plus the Blueprint Designer.

Infrastructure

The Infrastructure tab contains the bulk of items relating to the configuration of vRA 8. The first item that most vRA administrators will have to head to is Cloud Accounts, under the Connections heading. This is where the account details for various public cloud and VMware offerings are configured. A typical scenario in this area could be vCenter and a couple of public clouds configured here.

Cloud Accounts already configured for vRealize Automation
Cloud Accounts already configured

The other item under Connections is Integrations, where a range of VMware and third party integrations can be added. By default, there will already be an entry for the embedded vRealize Orchestrator (vRO) instance.

Integration Options for vRealize Automation
Integration Options

The next section of interest is Configure, which contains the items of Projects, Cloud Zones, Kubernetes Zones, Flavor Mappings, Image Mappings, Network Profiles, Storage Profiles and Tags. Some of these are mentioned during the Guided Setup (as shown in the Getting Started post).

Network Profiles allow the creation of objects that control network behaviour and settings. The options that become available when creating a Network Profile depend on the Cloud Account selected. For AWS, the settings include the ability to create on-demand networks or security groups, and the selection of existing networks. Tags can also be applied. For vCenter-based Network Profiles, there are options to add IP ranges. When selecting existing networks to use, discovered items are shown with extra information. For AWS, this can include the CIDR or whether public IPs are enabled.

Storage Profiles control the way storage is provisioned for virtual machines. For vCenter Cloud Accounts, the standard set of settings are exposed, such as Storage policy, thin/thick provisioning, and datastore. For other Cloud Account types, the expected options are exposed.

Storage Profile options for AWS
Storage Profile options for AWS

The Tags item lists all the tags discovered across all the Cloud Accounts. It’s then possible to select one or more tags and see what objects are currently assigned those tags. A good use case for this would be if a cost center tagging system was being used. It would be possible to see what resources a particular cost center is using across all platforms.

Tag Usage by the Windows Infrastructure Operations Team
Tag Usage by the Windows Infrastructure Operations Team

The Resources section sits under Configure and has a by-type breakdown of all resource items that vRealize Automation can see. These items are Compute, Networks, Security, Storage, Machines, Volumes and Kubernetes. Compute lists high level abstractions of the compute capability in each Cloud Account and the administrator can’t do much in here except apply tags.

The Networks item has more content and actions available. It lists all networks discovered across Cloud Accounts. There’s also the ability to manage IP ranges, review individual IP addresses being managed, view load balancers and view “network domains”. These network domains are the top-level network object for each platform type (ie. for AWS, that would be VPCs).

The Security item lists only Security Groups that have been discovered. In the case of my configuration, this meant security groups from AWS. The only action that can be performed in this area is to add or remove tags.

Storage has three tabs – Storage Policies, Datastores/Clusters and Storage Accounts. The first two tabs are vCenter-focused, with Storage Policies listed the discovered policies. Tags can be managed on these policies. Datastores/clusters is has similar functionality, appearing to be mainly for informational and tagging purposes. Storage Accounts lists any defined Azure storage accounts and can be tagged.

The Machines item lists all the machines across Cloud Accounts and includes information like status, IP address, Project, Owner and tags. There’s also the ability to filter the list by a variety of criteria.

The Machines list
The Machines list

Volumes lists all the “volumes” that have been discovered, which seems to include CD-ROM and floppy disk drives on vCenter VMs. Unfortunately the interface doesn’t list what machine the volume is attached to in the list view. It is possible to find this information by drilling down into the object.

The last area of any real interest under the Infrastructure tab is Onboarding. This area relates to creating “onboarding plans” for machines the vRA has discovered but isn’t managing. A benefit of this onboarding process is that vRA will create a blueprint based on the machines imported.

Extensibility

The Extensibility tab appears to share a lot of common elements with vRA 7, allowing the use of “Subscriptions” for triggering Orchestrator workflows. The menu items for this tab are shown below.

Extensibility menu items
Extensibility menu items

The Workflows item lists the 463 vRealize Orchestrator workflows that come with version 8, but it doesn’t seem to be anything to do in this area except look at them. The Actions item has no items listed, even though Orchestrator has over 400. Creating a new Action loads a code editor where the administrator can opt to write their code. By default, it seems to load the “Custom script” template which exposes the new ability to write python or nodejs code.

Action code editor
Action code editor

These Python-based scripts are part of the new “Action-based Extensibility” (ABX) that VMware have introduced in vRA 8. They are similar to how one might use vRealize Orchestrator workflows in vRA – to have automated tasks run at specific triggers. There is a reference in the documentation on ABX that suggests the code is actually run in the cloud, specifically using AWS Lambda. As such, an AWS subscription is required.

Marketplace

The Marketplace tab is an extended version of that which appears in Lifecycle Manager (LCM). It has three main areas – Blueprints, Images and Downloads. Images is a like-for-like match to LCM’s Marketplace content, containing virtual appliances from Bitnami and other vendors.

Marketplace Images
Marketplace Images

The Blueprints section is the more traditional blueprint items. Some are clearly designed for cloud-based deployments, such as one that uses AWS Redshift. Others are VMware-focused, relying on NSX. At the time of writing, only 18 blueprints are available.

Blueprints

The Blueprints tab represents a major area of use for infrastructure engineers and developers. This is where blueprints can be created, tested and reviewed. At a high level, the Blueprint Designer in version 8 has a lot of common elements with version 7.

vRealize Automation 8 Blueprint Designer
vRA 8 Blueprint Designer

Two major changes are visible in the screenshot above. Firstly, with the Infrastructure-as-Code (IAC) approach in vRA 8, the actual code of the blueprint is shown on the right side. A lot of the detail defining in a blueprint is now performed in this code area, as opposed to GUI elements in version 7. The other major shift is the expanded range of components from cloud services. It’s now possible to have cloud services like AWS’s RDS or Lamba or Azure’s Key Vault directly on the blueprint.

Verison 7 had a nested approach where certain objects could be nested inside of others. This seems to be gone now in version 8, with objects related to each other. A good example of this is the Configuration Management items like Puppet and Ansible. Adding items to the blueprint will generate a skeleton of code on the right side.

Generated code for a vSphere Machine component
Generated code for a vSphere Machine component

In some cases, a lightbulb will appear which assists in filling out the code. In the case of the vSphere machine, clicking this icon displayed a list of optional parameters that could be added. When using certain blueprint items, the fields that require values will show available options when clicking inside the quotes. For example, when clicking in the image or flavor properties for a Cloud Agnostic VM, Image and Flavor Mapping items will be listed.

Available values for Flavor
Available values for Flavor

The code view also has real time syntax checking to point out errors. This helps ensure the code is valid and will work when deployed. It’s possible to define a number of inputs for a blueprint, removing the need to hardcode a lot of values. There’s a decent range of controls that can be applied to the inputs to prevent bad values being entered.

Defining a "count" input and binding the value
Defining a “count” input and binding the value

With the Infrastructure as Code focus, blueprints now have built in versioning support. This helps avoid the sort of thing I’ve seen in some organisations where version control is done by copying the blueprint repeatedly with an incrementing number in the name. With this proper versioning capability comes support tools like code diff.

Blueprint diff
Blueprint diff

Closing Thoughts

Now that I’m getting into the actual areas of day-to-day use of vRealize Automation 8, it’s clear that in some areas there’s a lot that’s changed under the hood and for the better. The versioning support is a good acknowledgement that there’s often more than just one vRA administrator in many organisations and it helps avoid that “stepping on each other’s toes” situation. The code-based approach for blueprints may be a change for some. Those who have already come from IAC backgrounds (especially AWS Cloudformation or Terraform) should feel very comfortable with the concepts in the code-based blueprints.

vRealize Automation 8 First Impressions – Getting Started

Initial Login and Quickstart

Towards the end of the Easy Install wizard for vRealize Automation 8 (vRA 8), a link is provided for the vRealize Automation 8 UI. Accessing this link will load a landing page that shows a short piece of text and a link to a login button.

vRealize Automation Landing Page
vRealize Automation Landing Page

Read more

vRealize Automation 8 First Impressions – Lifecycle Manager

vRealize Lifecycle Manager (LCM) is the first component installed by vRA 8’s unified “Easy Installer”. One of its primary functions is the deployment of VMware’s vRealize products. As mentioned in my Installation post, a link to the LCM UI appears towards the end of the installation process. The Dashboard of LCM has five items: Lifecycle Operations, Locker, User Management, Content Management and Marketplace

Read more

vRealize Automation 8 First Impressions – Installation

vRealize Automation 7 has been travelling along for a while now. While it’s now at a level of maturity, it’s always been a complicated application, even just in terms of infrastructure (with the need for Windows-based “IAAS” servers). vRealize Automation 8 would appear to represent a tipping point for a lot of things VMware has been working on in the background across multiple products and technologies.

The deployment architecture in vRA 8 represents a significant shift from prior versions. The installer deploys three virtual machines. These VMs will consume a total resource set of 12 vCPU, 44GB of RAM and about 246GB of disk space.

Read more

Managing Local Admins via vRealize Automation

One of the major benefits of vRealize Automation (vRA) is the ability to add and extend the “Actions” available. These Actions enable self-service by the customer. One scenario I wanted to try was allowing someone to manage local administrators on a virtual machine they had provisioned.

Creating The Workflow

The starting point with this is creating a Workflow in vRealize Orchestrator (vRO). Managing local administrators would mean being able to add and remove members, so if I wanted it as a single workflow, there would be some sort of branching logic, such as the flowchart below:

Simple Workflow

Read more

Cumulative Update for vRealize Automation 7.6

UPDATE: Since I wrote this post, VMware have taken down this update because it causes issues with multi-tab XAAS forms. Specifically, it appears that each tab of a XAAS form will submit a request. One of these requests is “real” and will process properly, while the others will error out. As of 24th October, the hotfix is still offline. The issue with Chrome not rendering buttons was resolved in Chrome 76.

Earlier this month, VMware released the first Cumulative Update for vRealize Automation 7.6. The knowledge base article for it is available at https://kb.vmware.com/s/article/70911 The patch file is about 1.1GB.

The first resolved issue in the article is most likely the one that people are experiencing and the easiest to see – “Deployment forms are missing the Submit, Next and Cancel buttons” when viewed in Chrome 75. Below is a comparison of how the form is rendered in Chrome 75 versus Microsoft Edge.

Chrome 75 rendering Microsoft Edge Rendering

This issue by itself is a big showstopper if an organisation uses this functionality. I suspect this update will be a “must have” for them.

Read more

Thoughts on User Experience and Automation

The title suggests an odd combination of topics. I believe there still is an aspect of User Experience in IT Automation. With Automation, a task that was previously performed by a person is being done with a script or the like. This is really one half of the scenario, as we are also replacing the interface our customer was using and putting a new one in place. In some cases, this new interface can be worse than what was in place before.

I’ve collected a number of observations and thoughts about this User Experience aspect and I’ll go through some of them in this article. The examples I present will be relating to PowerShell and vRealize Orchestrator since those are tools I use frequently. But they could probably apply to any language/technology scenario involving automation.

Using The Appropriate Input Channel

This is a “sin” I see often with some people when they first get into PowerShell. I suspect this sin occurs because they were thinking in the context of their script being executed interactively by the user. This leads them to use the cmdlet Read-Host to get input from the user. In a way, Read-Host is seductive in its use. It allows you to assign the user’s input to a variable. The prompt the user sees can also be customised.

Using Read-Host to ask for input
Using Read-Host to ask for a username

The problem with this approach is the script will wait for the user’s input. This makes it unsuitable for scenarios where the script would be programmatically executed. In the case of PowerShell specifically, using Read-Host instead of Parameters seems counter-intuitive to the Pipeline model that PowerShell uses, where objects can be easily passed from one command to another. Lastly, Read-Host has no validation of any sort, so performing validation of the input would require extra code. Which leads into the next item…

Using Input Validation

Input Validation can help serve two goals. Firstly providing immediate feedback to the consumer of your automation that something bad was entered. Secondly protecting your automation against undesirable inputs. A bad example the first goal: a coworker who was trying to update details on some Virtual Machines, which involved inputting a project code. He had entered a value that the system didn’t like, but only informed him after he had submitted the request. This created a poor user experience because he didn’t get immediate feedback. The result was he had to fill in and submit the form again.

In the story about the co-worker the form was the front-end of a workflow in vRealize Orchestrator. Assuming the project code was always going to be a number, this issue could’ve been fixed by setting the data type of the input to “number”. A better approach would’ve been to use data binding and bind a list of valid project codes to the input.

Number validation in vRealize Orchestrator
vRealize Orchestrator’s number validation

PowerShell has a good range of options for validating input as well, including regular expressions.

Expose Complexity When Required

I consider the wizard for creating an EC2 instance in AWS to be a good benchmark for this. It’s possible to only select the “What” of the instance (what it does). After this, the consumer can immediately select “Review and Launch” from the instance type screen. If they want more control over the details, then they can take a different path. However, taking the “express path” requires very little input or technical knowledge to provision the instance.

A bad example of this: a provisioning workflow prompted for about 15 pieces of information, including which data center the server would live in, which storage, what network zone and so on. In this case, an overwhelming amount of complexity was being exposed by default. It required the consumer of this workflow to have intimate knowledge of the organisation’s infrastructure. This immediately restricts the potential scope of consumers for this workflow, and thus reduces the benefits it would realise.

A middle ground example was a private cloud system I supported, that used similar technology to the bad example above. Consumers could provision a virtual machine with a few abstract pieces of information (what is the server’s role/job? Is it a dev, test or prod system?). However, under certain conditions, like the server’s home being production, more complexity was exposed. The consumer was prompted for things like backup settings. In this case, more complexity was exposed in a way that was appropriate.

Closing Thoughts

There are many more problematic user experience scenarios outside the ones I’ve mentioned. These issues can be negated with some consideration on what is presented to them. This results in a positive experience of those consumers. A positive experience means they’re likely to use it more often and that’s a good outcome.

vCenter 6.7 Update 2 – Code Capture

One of the very cool new features that came with the latest vCenter update is Code Capture. This feature allows you to “record” actions in the HTML 5 web client. When the “recording” is ended, Code Capture will generate PowerCLI.NET code. For those who used Exchange 2007, you may remember a similar feature in the GUI management console. At the end of each wizard, there was a summary of PowerShell code that would perform the same task you just finished.

Code Capture has been around for a while as part of the HTML 5 Client Web Fling. However, at the time, it was not appropriate for production use. With the feature being included in 6.7 Update 2, it now is appropriate for use.

Read more

VMware vRealize Automation 7.6 – What’s New

VMware released an update for vRealize Automation (vRA) at the start of this month. This was a slight increment from 7.5 to 7.6. At a high level, two key areas of change are NSX integration and the vRealize Orchestrator (vRO) user experience. The Release Notes goes into a bit more detail and I’ll be using those more detailed items as a guide for the content in this post. I will be skipping over the new NSX pieces as I don’t have those available to me.

Read more

vRealize Automation – Dealing with a disconnected/orphaned IAAS Server

Following the installation of a replacement UPS I made a horrible discovery – there was problems with my vRealize Automation setup. The first sign of what was to come was when I reviewed the console of the vRealize Automation (vRA) Virtual Applications.

Things have gone badly…

Anyone who has had to setup vRealize Automation would probably know the sinking feeling I started feeling. Having to setup vRA again from scratch was not something I really wanted to do….

Read more