frankdenneman.nl

This year VMworld provided a broad selection of talks focusing on various forms of Kubernetes. Which is not surprising at all. Many organizations move away from buying and installing shrink-wrapped software and move towards in-house built custom applications. And what is the modern developer tool of choice? For many, it is the container. It’s expected to have 1.5 Billion containers shipped by the end of 2021.

Containers are nothing more than a new format of virtualized workload. Michael Gasch explains it very well in our session Deep Dive: The Value of Running Kubernetes on vSphere (CNA1553BU), where containers are task structs in the Linux kernel, not very different than executing an LS command. Well, a bit more than that as containers require CPU, memory, network, storage, and security.
Containers satisfy the developers’ need for speed, and they remove dependencies on underlying operating systems. When deploying massive amounts of containers, you need a container management platform, and Kubernetes is clearly the defacto standard in the industry.

Source: Cloud Native Computing Foundation
For the infrastructure team, running Kubernetes can provide a way to create an infrastructure agnostic platform. That is, it can run on any cloud. VMware is fully vested in making this happen; you can run containers natively (VIC), containers and Kubernetes in Linux VMs on vSphere. Pivotal Container Service (PKS) on-prem or in-cloud that helps customer deploy and operationalize day 1 and day 2 kubernetes solution and VMware Kubernetes Engine (VKE) (Kubernetes as a Service) for organizations who want to consume Kubernetes without owning, building or operationalizing any infrastructure.

I’ve selected a few VMworld sessions that cover these container consumption models. There are many more, and please check them out at the VMworld On-Demand Video Library.


Container and Kubernetes 101 for vSphere Admins (CNA1564BU)
A very popular session at VMworld was the 101 session for vSphere Admins. Nathan Ness and Sachin Thatte go over the basics of Container, Kubernetes and Pivotal Container Services. A very helpful primer for the rest of the listed videos. (Watch here)



Running Kubernetes on vSphere Deep Dive: The Value of Running Kubernetes on vSphere (CNA1553BU)
Michael Gasch (Resident Kubernetes Expert at VMware) and I go over the reasons why vSphere and Kubernetes are better together. We provide guidelines on how to successfully run your Kubernetes environment.
(Watch Here)



A Deep Dive on Why Storage Matters in a Cloud-Native World (HCI1813BU)
7 out of 10 applications that run in containers are stateful applications (source: Datadog), you want to provide persistent storage. Myles and Tushar talk about project Hatchway and provide a preview of the upcoming Cloud Native Storage (CNS) Control plane.
(Watch Here)



Operating and Managing Kubernetes on Day 2 with PKS (CNA1075BU)
If you are planning to run large-scale kubernetes deployments on-prem, you should consider Pivotal Container Service (PKS). PKS allows you to deploy multiple kubernetes clusters quite easily. Thomas Kraus and Merlin Glynn show how to tackle day 2 operations and review SDDC products, such as vRealize and Wavefront, that integrates with PKS.
(Watch Here)



VMware Kubernetes Engine
VMware Kubernetes Engine (VKE) offers a turn-key solution of managed Kubernetes clusters that run natively on AWS. Not in VMware Cloud on AWS, not on vSphere, pure native EC2! Plans are to run VKE at multiple cloud providers, allowing you to create environments that no-other cloud provider themselves can provide. Think about an HA cluster spanning both AWS and Azure. However, we are not that far right now, but it is interesting to take a look at what VKE is and how Smart Clusters will change the way you will operate Kubernetes.
Intro to VMware Kubernetes Engine-Managed K8s Service on Public Cloud (CNA2084BU)
Tom and Valentina go over the concepts and customer value of VKE, including a nice demo. (Watch Here)



Deep Dive: VMware Kubernetes Engine-K8s as a Service on Public Cloud (CNA3124BU)
After getting familiar with VKE, I recommend to watch the session of Tom and Alain. They dive deeper into the concept of Smart Clusters. (Watch Here)

I hope you enjoy watching these sessions, please leave a comment about sessions you think are worth watching.

During VMworld, the DRS performance team released a new tech paper covering the DRS Enhancements in vSphere 6.7. It’s a short white paper uncovering the interesting improvements made to DRS. Download it here.

The new version of the vSphere clustering deep dive is available on Amazon. The vSphere 6.7 Clustering Deep Dive is the fourth edition of the best selling series. Over 50.000 clustering deep dive books have been distributed, and I hope this version will find its way on your desk.

The new version of the clustering deep dive covers HA, DRS, Storage DRS, Storage I/O Control and Network I/O Control. In the last part of the book, we bring all the theory together and apply it to create and describe a stretched cluster configuration.
Now, why am I using the title vSphere 6.x Deep Dive Resource Kit? Well, it’s because we believe that when you pair this with the vSphere 6.5 Host Resource Deep Dive book, you get this bundle that allows you to understand the core of your virtual infrastructure.

Changing the Game
When Duncan and I set out to write the 4.1 HA and DRS deep dive, we wanted to change the content of technical books. Instead of having a collection of screenshots paired with the text, next, next finish, we wanted to provide a thorough explanation of what happens under the cover. When you push this button, this happens in the code. By uncovering the inside, we arm the administrator and architect with the knowledge to create or troubleshoot any architecture anywhere.

When combining these books together, it creates a real end-to-end guide for your architecture. For example, in the DRS section, we explain how the cluster determines the resource entitlement of the VMs in a resource pool. In the vSphere 6.5 Host resource deep dive, we describe the inner workings of the memory and CPU scheduler and how they allocate the physical resources based on the resource entitlement of the VM.
Back Side of the Book
When releasing the host resource deep dive, we came up with a cool little logo of a divers helmet. If you want to get deep, you need more than a snorkel. One divers helmet to explore the host, but in the cluster deep dive, we cover multiple hosts, grouped in a cluster. What do you need when you need a lot of people to explore the deep? You need a submarine! 😉 It might even end up on some T-shirt.

New Name on the Cover
As you might have noticed, a new name appears on the cover. We asked Niels Hagoort to help us to cover the quality of service aspect of the book. Niels dove into the deeps of Storage I/O Control and Network I/O Control and created an excellent addition to the book.

Foreword
And last but not least, the foreword. In the previous books, industry luminaries generously provided us with amazing forewords. This time we looked at the community. We asked Chris Wahl to write the introduction. Chris has been an early supporter of the book series, and he has helped the community in many ways. We asked him to provide us with his point of view.
I hope you enjoy the book as much as we enjoyed writing it.

Yesterday, Kenneth Hui was on stage at the VTUG providing his personal opinion about VMware Cloud on AWS. The reason I say personal is that he forgot to remove the Rubrik Logo’s from his slide (I checked with Rubrik).

On one slide he mentions that the SDDC, that is the Software Defined Data Center provided by VMware Cloud on AWS (VMC) is not an SDDC out of the box. And to me, that sounds a bit weird. Let’s go over the process of spinning up an SDDC.
First, you log onto vmc.vmware.com and you sign up for the service. In the console you define the number of hosts for deployment, click apply.

If you select a multi-host deployment, by default an SDDC cluster contains 4 hosts, that means that VMC deploys four physical hosts (for more info: Dedicated Hardware in a Public Cloud World) on the AWS infrastructure. It installs and configures vSphere, vSAN, and NSX for you automatically. After roughly two hours you are the sole-owner of dedicated hardware with a fully software-defined data center running on top of that. Just log into your in-cloud vCenter and start to deploy your workload. So to reiterate, you just clicked a button on a website and a fully functional data center is deployed for you.

From an initial deployment standpoint but I was referring to day 2 operations.

— Kenneth Hui (@kenhuiny) July 19, 2018

Ok so what about day 2 operations, let’s define this a bit clearer because there are multiple definitions available. Dzone provides the following definition: Once “something” goes into operations, “day 2 operations” is the remaining time period until this “something” isn’t killed or replaced with “something else.”
We build a cloud management platform in AWS in order to deal with day-2 operations. VMware provides the service, we will keep the lights on for you, troubleshoot and maintain your environment. This CMP plaform allows us to provide services like automated hardware remediation. If a component inside the ESXi hosts fails, such as a NIC, or an NVMe device, the backend will detect this and it will initiate a process to replace the faulty host with a fully operational one. The customer won’t have to do a thing.

Elastic DRS allows the cluster to respond to workload utilization automatically. It allows for automatic scale-out and scale-in, without the need for human intervention.
Stretched Clusters protects the workload in the Cloud SDDC from AZ outages. If something happens, HA detects the failed VMs and restarts them on different physical servers in the remaining AZ without manual human involvement.
Content library, allows the customer to subscribe the in-cloud SDDC to a template repository that automatically provides VM templates to the in-cloud SDDC. Read Williams post for more info
Disaster Recovery as a Service, just go to the console, enable the add-on and the in-cloud components for SRM and vSphere Replication are automatically deployed and configured. Connect it to your on-prem components and you can build your DR runbooks.
And there are many more functions that cover the lights-on, maintenance, housekeeping and optimize tasks of day 2.
Now with that explained, the stories continue and a debate broke out on twitter. Some said it needs a form of CMP (eg. vRealize) for operating the SDDC.

Depends on your definition of SDDC, doesn't it?

From the DMTF definition: "resources are dynamically discovered, provisioned, and configured based on workload requirements."

Sounds like a CMP is needed to automate provisioning of resources according to that definition.

— Ken Nalbone (@KenNalbone) July 20, 2018

This is an interesting observation, for which operation? Not for life-cycle or infrastructure management. We will take care of that for you. VMC is a fully managed service by VMware. It is responsible for the uptime and the lifecycle of the SDDC.
we have built a CMP platform on the AWS infrastructure that allows us to deal with VMC. In a presentation of Chris Wegner (one of the principal engineers of VMC) the architecture is explained.

The blue box is the actual SDDC, The green box is a custom-built CMP that allows VMware to identify customers, billing customers, providing support for customers (as a VMC customer, you only deal with VMware) but most importantly for this story, it allows VMware to deploy hardware and software (Fleet management). The next image provides a more detailed view of the green box.

This is what you need to support hundreds of SDDCs across multiple regions (Oregon, N. Virginia, London, Frankfurt). Here you can see the bits for provisioning management, dealing with AWS services, acquiring hardware, configuring all the software and of course the ability to troubleshoot.
You as a customer, do not need to worry again about ripping and replacing hardware because it failed, or because it’s nearing the end of support. You only need to care about deploying your workload. And because we took the conscious decision of using vCenter as the management structure, you can use your on-prem vRealize suite and deploy your workload on-prem or in-cloud. Using vRealize to deploy workload is the way to go forward because 80% of our customers have a hybrid cloud strategy a on-prem deployment is expected. It makes sense to run your tooling on-premises.
With VMware Cloud on AWS, your responsibility shift from managing hardware to managing the consumption of resources.