Category Archives: Cloud

Putting your Cloud on Autopilot and #CloudLife

The Conference

On June 23rd, I was delighted to speak for the 3rd year at the Looking Ahead 2016 summit. I’ve talked about how much I love my job before and I can say that our summit reinforces that for me every single year. I leave feeling energized as we take risk after risk every year and try to show customers where we are heading and how we can improve their lives.

My Session: Putting Your Cloud on Autopilot

First of all, I would absolutely love feedback on the session, so please send me an e-mail or tweet me. I really appreciate it.

Approaching this years session, it was clear to me so many of the customers I deal with on a day to day basis have moved beyond what I often call the “plumbing phase” of Cloud. I decided to reinforce the message around Cloud by starting off with what it means to me and the Ahead team in general. I am fairly sure every session I do on Cloud for the rest of my life will start off with 60 seconds of what we exactly mean by it; given how misused the term is in the industry.

Deployment Models

Once we got over the basics of doing Infrastructure as a Service, it was time to move onto newer items. In the past I’ve talked a lot about Self Healing Datacenter and how to actually make that a reality, but this time I wanted to focus on the different ways Automation can help across On-Premises and the Public Cloud.

Essentially going from the IaaS Approach via Puppet…

To a partial refactor using AWS RDS…

To a complete PaaS deployment…

All using the same application. I completed a demo showing this, as well as the various ways AWS failover works. The main point here is to stress the choice and flexibility you give up by embracing the various deployment models. I remember saying a few years ago “No 2 clouds are the same”, and that seems to have taken off. I think it’s still valid, at least for now.

Self Healing

Then it was time to get back onto the Autopilot theme again, this time using a Google Car to illustrate the mechanisms we use in the real world to create safety. Relating it back to Cloud, I explained an example of event management using AWS Lambda and ServiceNow. I took an AWS Lambda function and used it to connected to ServiceNow so as nodes spun up or spun down ServiceNow Change records would be created automatically. I’ve got a post brewing on the benefits of Orchestration and Event Driven Automation which I hope to finish up some time. I think this is a key topic, often overlooked these days and something I’ve been discussing heavily with our team at Ahead.

Finally – The Cloud Experience

If there’s one thing I get fed up with at VMUGs and other user groups, it’s people standing up and saying you need to program and that’s the skill. While important, I feel like many just state the obvious in career development without truly explaining what it means to have a functioning Cloud and how you get to that, across On-Premises and in the Datacenter.

Nick Rodriguez and I came up with a new term which we call #CloudLife (Also a future blog post). How do you create the awesome experience that truly changes behaviours in an orgnaization? I talk also with my colleague, Dave Janusz, on this topic alone at length. How do you make someone do something in your IT environment without having to tell them? I love asking this question as it creates all sorts of interesting ideas for design best practices. I’m going to write more on this topic soon also but I hope people start to realize the most successful clouds are the ones that create a user experience that works. I read a book during my University days when I studied a module on Human Computer Interaction. I still state to this day, that the book I read combined with the module taught me some of the most important lessons in IT.

If you haven’t got it, check it out below. It’s a fun read and not entirely related to IT, but I loved it:

<br />

Remember, programming is important, but it’s not the only major skill.

With that, I’m going to end this post. I hope to finally sit down soon and write 3 posts I’ve been thinking and talking about for a while…

What is #CloudLife?
Skills of Successful Cloud Deployments
Visual Orchestration vs Non-Visual Orchestration

These topics deserve more debate than they get today. I feel like the DevOps initiatives when done as a Silo (yup you heard me, people do DevOps in a silo that they call DevOps), have masked some of the changes IT has to make. Also IT hasn’t always been able to articulate and truly create the services Developers always needPublic Cloud is here, but there’s more to wrap around it. Do Developers use visual studio and connect directly to Azure? Do they use Docker + IaaS for more flexibility? How do you present the right services and lego bricks of automation?

Time to dream more about….#CloudLife

A ninja walks into a #TurboFest

*Very late posting this, but better late than never!*

On Jun 15th I was fortunate enough to attend my first VMTurbo TurboFest. VMTurbo is a company I’ve personally been watching for a while, and have always been excited to see the direction they are taking, not just in the Datacenter, but Cloud operations in general.

On top of that, Amy, along with her director at UCMC Jason Cherry, were attending in order to present their results with VMTurbo as well as the new wave of Automation Amy is leading over there.

About the day

What stood out for me most, was just how informal and social the event was. I thought this was a nice touch compared to other events. It was also extremely customer focused, something I think many other events lose.

The kickoff was great as we learned from CEO Benjamin Nye the direction VMTurbo continues to focus in.

VMTurbo truly GET that Cloud is real and they are working quickly to adapt to the increased momentum of Public Cloud adoption. VMTurbo treats everything as a commodity and works to create a level of abstraction and allow automated intelligence to determine workload placement.

Given how many vendors have done a poor job at trying to offer automated decisions based on Public Cloud models, I’m excited to see another company with a proven track record in this space on-premises, fully embracing this gap in the market.

The UCMC Presentation

One of the main reasons I attended was to support my fellow ninja blogger as well as the UCMC team who I’ve had the pleasure of working with.

What strikes me most in all my talks with Amy and Jason is the results they’ve been able to achieve. $600,000 in cost avoidance due to properly utilizing their VMware environments. They saw instant benefits by adding VMTurbo to their infrastructure management tools and were able to get higher density on their clusters as a direct results of the VMTurbo software.

Amy went on to talk about the work around the UCMC Cloud they have been building, primarily on-premises now, with future expansion to Public.

UCMC will be using a mixture of ServiceNow, vRealize Orchestrator and Puppet to easily automate their deployments and take UCMC IT into the 21st century as they put it. This is similar to the solution we both worked on during our time at a previous employer, and the methodology is clean and simple to use.

In addition, Amy has been developing a number of vRO Workflows around the VMTurbo plugin specifically for workload placement and chargeback. This placement workflow is called before the VM clone workflow is initiated in order to determine the optimal host location of it. Handy way to integrate the placement decision directly into the workflow instead of having to go to VMTurbo first, get the answer, and then manually key it in as an input. Awesome stuff.

Stay tuned as Amy plans to release all the workflows on our blog when finished.

After the community sessions, the panel sat back down and had a very lively Q&A session with lots of users in the community asking great questions.

Before I left, the VMTurbo marketing team played one of the most impressive videos I’ve honestly ever seen. I’m trying to get a youtube link of it to post here as it is pure awesome. Think VMs meet insane action movie with awesome effects.

With that said, I can’t wait to do more with VMTurbo again myself, and look forward to catching up with the team again at VMworld.

IaaS Fundamentals: Creating a fresh Windows Server 2012 Template – Part 2

With our base VMware vSphere VM shell ready, it’s time to continue installing the Windows OS.

Just before we dive in, it is worth noting that depending on how you are remotely connected into the desktop, you may have issues controlling your mouse. In my case I was going via a View Desktop and then into the VRM console. I decided to just use the Tab and Spacebar key instead to make my selections. This will get much easier later on when VMware Tools is installed in the VM.

Select Install now, accept the defaults for language etc. until you get to the type of OS you wish to deploy.
I choose the Datacenter Edition with GUI here. Note: You can always remove the GUI and go back to ServerCore if needed. I know in my environment our Windows team still generally uses the GUI
Click Next once chosen

Accept the license terms and click Next
Change the installation type to Custom: Install Windows Only (advanced) and click Next.

Next you will be prompted for your drive layout. It should look like the screenshot below unless you chose a different drive configuration.

Leave Drive 0 selected and click Next

Sit back relax and enjoy the show!

Enjoy some tea while you wait…

Once finished you will need to enter your Administrator Password for your Windows Template.

Coming soon – Part 3 – Configuring and tuning your OS

Rubrik Announces r528 Cloud Appliance and Sexy New Features

Rubrik announced the r528 cloud appliance today. Yes, Rubrik just got sexier. Not only has Rubrik grown exponentially as a company ,they are on their 3rd update and are now quite the global force with 90+ signed Channel Partners and 4PB+ Protected in the Field

The r528 offers encryption at rest and in flight from VMware. Because the appliance is using hardware encryption, there is no compromise on speed or performance. The self encrypting drives (SED), use AES 256 circuitry. All data written to disk is encrypted automatically and data read is decrypted automatically. Eliminating or overwriting the security key would perform an instantaneous wipe. If a drive were to be taken out, it would be deemed worthless without the key

Boring stuff you should know about: NIST

This offering is FIPS 140-2 Security validated. What does that mean? It sounds important. The drives and Rubrik Cryptographic library are FIPS 140-2 certified. Where most backup appliances are Level 1, Level 2 brings about the ability to detect physical tampering. If you want to nerd out and read up on FIPS 140 here. From there, you can read that FIPS 140-2 Level 1 provides the lowest level of security. Basic security requirements are specified for a cryptographic module (e.g., at least one Approved algorithm or Approved security function shall be used). No specific physical security mechanisms are required in a Security Level 1 cryptographic module beyond the basic requirement for production-grade components. FIPS 140-2 Level 2 improves upon the physical security mechanisms of a Security Level 1 cryptographic module by requiring features that show evidence of tampering, including tamper-evident coatings or seals that must be broken to attain physical access to the plaintext cryptographic keys and critical security parameters (CSPs) within the module, or pick-resistant locks on covers or doors to protect against unauthorized physical access.

For key management, Rubrik supports external key standards using KMIP 1.0 or they also provide a Trusted Platform Module (TPM) so there would be no need for KMS . Giving the customer options if they don’t have KMS setup in their environment.

But wait there’s more!

Rubrik Converged Data Management 2.2

Enhancing the auto protect and SLA inheritance Rubrik already offers

Dynamic Assignment – Set policy on a vCenter, Data Center, Cluster, Folder, Host, and more.
Inheritance Options – Any new object or workload created will automatically pick up parent SLA assignment.
Do Not Protect – Block SLA policy from being inherited with explicit denial to prevent data protection at any desired level.

Throttle detection!

Most people don’t want backups to affect workloads. The software can look for latency to make sure it’s not causing performance issues. If storage latency is rising, it is smart enough to halt additional tasks on the fly. Backups don’t continue to pile up on your environment like a WWE royal rumble.

Scalability

In testing, Rubrik has scaled out to a 20u, 10 brik, 40 node cluster. That’s insane
Protect 10,000 VMS using vSPhere 6.0
Instant recovery: Quicker spin up of clone workloads (thanks to being able to get 20,000 I/O per brik) and faster storage vMotion to your production environment.

Cluster Policy Enhancements

Global pause gives you the ability to use a maintenance window to perofrm work on the cluster
Recurring First Full Snapshot Window gives you control to say when a full backup should be performed within an SLA
New Retention Periods bring increased flexibility for SLA policies to meet different customer requirements
Blackout Windows define when no operational taks should be executed by the cluster

NAT Support

For customers that don’t want to use site-to-site tunneling , there is now NAT support for public bi-directional replications.

User Experience and Management

You might as well enjoy managing your backups …

Last but not least

Backup physical alongside your virtual environment. This includes SQL and Linux. You have all the capabilities for physical recovery that you are used to having for your VMs

Automation Fun

An oldie but goodie, you can tell from this blog, we’re all about automating all the things possible.

There are several options to satisfy the automation ninja within you:

GitHub PowerShell-Module Repository
PowerShell Gallery (NuGet)
Continuous Integration with AppVeyor

And a personal favorite:
vRealize Orchestrator Packages

Rubrik maintains it’s mantra don’t backup go forward with it’s continuous improvements to the backup experience.

Server Name Generator – Final

Now that we have our partial name creator and our workflow to check for the next available name, we can bring it all together for a complete server name workflow.

Complete Server Name Workflow

General Attributes:

partialName: Type = String

Inputs:

appType: Type = String
location: Type = String
network: Type = String
envLevel: Type = String
OS: Type = String
domainSuffix = string

Outputs:

vmName: Type = String
fqdnOut: Type = String

The Workflow:

As explained in Server Name Generator – Part 1, we create the partial name based on datacenter location, network, operating system and environment. We now append that with a number and check if the name exists in DNS using the worfklow from Server Name Generator – Part 2.

Now the output you will see is something similar to this in the logs:

[2016-04-18 15:01:08.930] [I] Entering Generate Partial Name Worfklow
[2016-04-18 15:01:08.935] [I] The datacenter shortname is AKL
[2016-04-18 15:01:08.948] [I] The network shortname is P1
[2016-04-18 15:01:08.981] [I] The OS shortname is W
[2016-04-18 15:01:08.994] [I] The environment shortname is 1
[2016-04-18 15:01:09.013] [I] The shortname of the application is APP
[2016-04-18 15:01:09.028] [I] The partial name of the VM is AKLP1W1APP
[2016-04-18 15:01:09.030] [I] Leaving Generate Partial Name Workflow
[2016-04-18 15:01:09.072] [I] Your partial name is: AKLP1W1APP
[2016-04-18 15:01:09.073] [I] Starting loop --- 
[2016-04-18 15:01:09.075] [I] Server Name: AKLP1W1APP01 found in master list - Incrementing with next number and starting over
[2016-04-18 15:01:09.078] [I] Server name: AKLP1W1APP02 not found in master list. Recording new name and continuing.
[2016-04-18 15:01:09.079] [I] Checking host for existing DNS record: AKLP1W1APP02.tritech.local
[2016-04-18 15:01:09.086] [I] Check complete for FQDN: AKLP1W1APP02.tritech.local Found IP address of: null
[2016-04-18 15:01:09.588] [I] 
[2016-04-18 15:01:09.591] [I] The vmName to pass as output is: AKLP1W1APP02
[2016-04-18 15:01:09.592] [I] The FQDN to use is: AKLP1W1APP02.tritech.local

AKLP1W1APP01 already existed so our loop went to the next number 02 and found that it didn’t exist.

This concludes the Server Naming series. You can download the complete workflow here. If you have any questions, or found this content useful, let us know with a comment.

Real World Cloud: Template Creation and Management

To bake in, or not to bake in, that is the question you must ask.

Keeping it simple…

Without a doubt, template creation and template sprawl are items people worry about when getting ready for their cloud environment. Maybe you have a Linux team that use PXE plus kickstart in the past and you’re having to convince them to move to templates, or maybe you had a bunch of templates with everything baked inside. When it comes to cloud, I’m personally a big believer in keeping your template as bare as possible, providing it doesn’t add a significant overhead to the build time.

This means stripping the template down to the base OS and layering everything on top.

The key also to remember here is that you want to try to be able to use a similar approach to your template management in your Public Cloud, as well as, your On-Premises Datacenter. This way you aren’t managing 2 completely separate processes and creating more complexity.

How do I configure things like agents, settings etc. after the template is cloned?

The rise of configuration management tools in the past few years has been key. Puppet and Chef in particular have become extremely widespread in the past few years. On top of that Microsoft is constantly evolving PowerShell DSC, and we have other newcomers like Saltstack and Ansible that are quickly gaining traction.

Essentially, everything you install and configure on top of the VM deployed, will be via Configuration Management. This allows us to completely abstract the configuration from the virtual machine. Think of it like a container (not to be confused with docker), that just hosts a specific configuration. The more you can modularize the pieces that sit on top of the OS, the easier it is to manage. This does however come at a cost of speed, because each application and potentially a number of reboots will need to be configured post VM provisioning.

How are Configuration Management Tools Changing?

I’m fortunate enough to work with some great Puppet engineers at Ahead who work on these pieces daily. Rather than retype what one of my colleagues at Ahead has already written, I suggest you check out this blog article on the Ahead Blog. http://blog.thinkahead.com/the-present-and-future-of-configuration-management

Creating the Template:

There are a number of ways to do this, and i’m sure the are more than the ones I list here:

vSphere: Clone from an Existing VM and prep it.
vSphere: Create a fresh template from an ISO.
Amazon: Create a Windows AMI from a running instance. (Use a prebuilt Amazon AMI basically and customize it)
Amazon: Import your vSphere VM and convert.

In the next post, we will be drilling down in detail on creating a fresh Windows vSphere Template for use in your Cloud. As we progress to Public Cloud, we will come back to the Amazon AMI options.