Multi-Cloud Load Balancing and Autoscaling with NSX Advanced Load Balancer (formerly Avi Networks)

Multi-Cloud Load Balancing and Autoscaling with NSX Advanced Load Balancer (formerly Avi Networks)

Do you want to build your private cloud like a hyperscaler is doing it? You know that VMware Cloud Foundation is becoming the new vSphere, but still wonder how you can implement software-defined load balancing (LB) or application services and features like autoscaling or predictive scaling? Then this article about multi-cloud load balancing and autoscaling with NSX Advanced Load Balancer aka Avi Networks is for you!

My Experience with a Legacy ADC

A few years ago, I was working on the customer side for an insurance company in Switzerland as a Citrix System Engineer. My daily tasks included the maintenance and operation of the Citrix environment, which included physical and virtual Citrix NetScaler Application Delivery Controller (ADC) appliances. The networking team owned a few hardware-based appliances (NetScaler SDX) with integrated virtualization capability (XenServer as hypervisor) to host multiple virtual NetScaler (VPX) instances.

The networking team had their dedicated NetScaler VPX instances (for LDAP and HTTP load balancing mostly) and deployed my appliances after I filed a change request. Today, you would call this multi-tenancy. For a Citrix architecture is was best practices to have one high availability (HA) pair for the internal and one HA pair for the external (DMZ) network access. A HA pair was running in a active/passive mode and I had to maintain the same setup for the test environment as well.

Since my virtual VPX appliances were hosted on the physical SDX appliance, I always relied on the network engineers, if I needed more resources (CPU, RAM, SSL, throughput) chips allocated to my virtual instances. Before I could upgrade to a specific firmware version, I also had to wait until they upgraded the physical NetScaler appliances and approved my change request. This meant, we had to plan changes and maintenance windows together and had to cross fingers, that their upgrade went well, that we could upgrade all our appliances after.

NetScaler SDX

It was also possible to download a VPX appliance, which could run on top of VMware vSphere. To be more independent, I decided to install four new VPX appliances (for the production and test environment) on vSphere and migrated the configuration from the appliances running on the physical SDX appliance.

Another experience I had with load balancers was when I started to work for Citrix as a consultant in Central Europe and had to perform a migration of physical NetScaler MPX appliances, which had no integrated virtualization capability. I believe I had two sites with each two of these powerful MPX appliances for tens of thousands of users. Beside the regular load balancing configuration for some of the Citrix components, I also had to configure Global Server Load Balancing (GSLB) in active/passive mode for the two sites.

NetScaler GSLB Active Passive

There were so many more features available (e.g. Web Application Firewall, Content Switching, Caching, Intrusion Detection), but I never used anything else than the NetScaler Universal Gateway for the remote access to the virtual desktop infrastructure (VDI), load balancing, HTTP to HTTPS redirections and GSLB. In all scenarios I had a HA pair where one instance was idling and doing nothing. And the active unit was in average not utilized more than 15-20%. It was common to install/buy too large or powerful instances/licenses, because you wanted to be on the safe side and have enough capacity to terminate all your SSL sessions and so on.

It (load balancing) was about distributing network traffic across multiple servers by spreading the requests and work evenly, and do add some intelligence (health monitoring) in case an application server or a service would fail or be unavailable for any reason. If one more application server was needed, I ordered a new Windows Server, installed and configured the Citrix components and added the necessary load balancing configuration on the NetScaler. These were all manual tasks. The same work has been done by the network engineers when the application team requested a new application server, which then had to be added to the load balancing configuration on their NetScaler appliances.

This was my personal experience from 2017. Since then applications became more complex and distributed. The analysts and market are focusing on containerized and portable apps running and more and more in multiple clouds. The prediction is also that the future is multi-cloud.

Multiple Clouds vs. Multi-Cloud

There are different definitions and understandings out there what multi-cloud means. In my understanding, using a private cloud, AWS, Microsoft Office 365 and Azure are a typical setup with multiple clouds. There are simple scenarios where you migrate workloads from the private to the public cloud (e.g. Azure) or having applications with services lying on the private on the public cloud. The latter would be an example of a hybrid cloud architecture.

The reasons for which services and resources are needed or distributed on multiple clouds (on-prem, Azure, AWS, GCP etc.) are various:

  • Avoid dependence on only one cloud provider
  • Consume different specific services that are not provided elsewhere
  • Optimize costs for different workloads and services
  • React to price changes by the providers

That is why we are seeing also the trend to break up big legacy applications (monoliths) in smaller pieces (segments), which is a best practice and design principle today. The goal is to move to a loosely coupled and more service-oriented architecture. This provides greater agility, more flexibility and easier scalability, because of less inter-dependencies.

And, if we take the second example from the list before, a segmented application is much easier to run in different clouds (portability). Running one application over multiple clouds is in my understanding the right definition of multi-cloud.

Multiple Clouds versus Multi-Cloud

Let’s assume that most probably all the four reasons above apply to larger enterprises. If we take another angle, we can define some business and technical requirements for multi-cloud:

  • Application or services need to be cloud vendor-agnostic
  • Provide or abstract control and management interface of multiple clouds
  • Support application portability/relocation between clouds
  • Combine IaaS and services from different clouds
  • Possibility to deploy components of applications in multiple clouds
  • (Cloud) Broker service needed
  • Policy and governance over multiple clouds
  • Network connectivity for migration scenarios with partially modernized applications
  • Automated procedures for deployments
  • Application monitoring over different clouds
  • Costs management
  • Lifecycle management of deployed applications in multiple clouds
  • Self-adaption and auto scaling features
  • Large team with various expertises needed

How can you deliver and manage the different applications services like load balancing, web application firewalling, analytics, automation and security over multiple clouds?

Another important question would be, how you want to manage the deployment on the various clouds. But cloud management or a cloud management platform is something for another article. 🙂

The requirements for the developers, operations and the business are very complex and it’s a long list (see above).

It is important, that you understand, based on the requirements for multi-cloud, that it is mandatory to implement a modern solution for your modernized application architectures. Enterprises have become more application-centric and everyone is talking about continious integration, continuous delivery and DevOps practices to automate operation and deployment tasks. A modern solution implicits a software-defined approach. Otherwise you won’t be able to be agile, adapt to changes and meet future requirements.

My past experience with Citrix’ NetScaler is a typical example that “virtualized” and “software-defined” are not the same thing. And this is very important if we want to have a future-ready solution. If we look at VMware’s software-defined data center (SDDC), beside the virtual compute, also includes software-defined storage and networking. Part of the software-defined networking portfolio is “NSX Advanced Load Balancer“, the software-defined application services platform, which was also known as “Avi Networks” before VMware bought them in June 2019.

Unlike a virtualized load-balancing appliance, a software-defined
application services platform separates the data and control planes
to deliver application services beyond load balancing, real-time
application analytics, security and monitoring, predictive autoscaling,
and end-to-end automation for Transport (Layer 4) to
Application (Layer 7) layer services. The platform supports multicloud
environments and provides software-defined application
services with infrastructure-agnostic deployments on bare metal
servers, virtual machines (VMs), and containers, in on-premises
data centers and private/public clouds.

Autoscaling became famous with AWS as it monitors your applications and automatically adjusts capacity to maintain availability and performance at the lowest possible costs. It automatically adds or removes application servers (e.g. EC2 instances), load balancers, applies the right network configuration and so on.

Can you achieve the same for your on-premises infrastructure with VMware? Yes.

Is there even a solution which can serve both worlds – on-prem and cloud? Yes.

And what about predictive scaling with real-time insights? Yes.

NSX Advanced Load Balancer (NSX ALB)

Why did VMware buy Avi? Because it follows the same architecture principles like NSX: A distributed platform with a separate control and data plane built on software-defined principles for any cloud.

Avi High Level Architecture

Traditional ADCs or load balancers are mostly configured in active/standby pairs, no matter if physical or virtual. Typically you would see around 15% utilization on the active node where the secondary standby node is just idling and doing nothing. Each pair is its own island of static capacity which shares the management, control and data plane.

You have to decide where to place the virtual IP (VIP) and how much you want to overprovision the physical or virtual appliances, because there is no capacity pooling available. This leads to operational complexity, especially when you have hundreds of such HA pairs running in different clouds. Therefore, legacy and virtualized ADCs are not the ideal choice for a multi-cloud architecture. Let’s check NSX ALB’s architecture:

Control Plane – This is the brain (single point of management) of the whole platform that can be spun up in your on-prem environment or in the cloud (also available as a managed SaaS offering), typically as a three-node cluster. Within this cluster, all configuration is done, this also where the policies reside and the decisions are made. It is the controller’s duty to place virtual services on SEs to load balance new applications or increase the capacity of running applications.

The control plane comprises the three pillars that deliver the key capabilities of the Avi platform:

  • Multi-Cloud – Consistent experience for any cloud, no lock-in
  • Intelligence – The machine learning based analytics engine enables application performance monitoring, troubleshooting, and operational insights (gathered by the SEs)
  • Automation – Elastic and predictive auto scaling & self-service without over-provisioning through a complete set of REST APIs

Data Plane –  The Service Engines (SEs) handle all data plane operations by receiving and executing instructions from the controller. The SEs perform load balancing and all client- and server-facing network interactions. It collects real-time application telemetry from application traffic flows. 

As already mentioned, NSX ALB can be deployed in multiple cloud environments like VMware vCenter, Amazon Web Services, Microsoft Azure, Google Cloud Platform, Oracle Cloud, IBM Cloud, VMC on AWS, Nutanix, OpenStack or bare-metal.

Use Cases

Most customers deploy Avi because of:

  • Load Balancer refresh
  • Multi-Cloud initiatives
  • Security including WAF, DDoS attack mitigation, achieve compliance (GDPR, PCI, HIPAA)
  • Container ingress (integrates via REST APIs with K8s ecosystems like GKE, PKS (TKGI), OpenShift, EKS, AKS, TKG)

Advanced Kubernetes Ingress Controller Avi Networks

  • Virtual Desktop Infrastructure (Citrix, VMware Horizon)

Consistent Application Services Platform (Features)

Avi/NSX ALB is an enterpise-grade solution. So, everything you would expect from a traditional ADC (e.g. F5), layer 4 to layer 7 services, SSL, DDoS, WAF etc. is built-in without the need for a special license edition. There is also no NSX license requirement even the product name would suggest it. It can be deployed as a standalone load balancer or as an integrated solution with other VMware products (e.g. VCF, vRA/vRO, Horizon, Tanzu etc.).

Avi Networks Features

Below is a list with the core features:

  • Enterprise-class load balancing – SSL termination, default gateway, GSLB, DNS, and other L4-L7 services
  • Multi-cloud load balancing – Intelligent traffic routing across multiple sites and across private or public clouds
  • Application performance monitoring – Monitor performance and record and replay network events like a Network DVR
  • Predictive autoscaling Application and load balancer scaling based on real-time traffic patterns
  • Self-service – For app developers with REST APIs to build services into applications
  • Cloud connectors – VMware Cloud on AWS, SDN/NFV controllers, OpenStack, AWS, GCP, Azure, Linux Server Cloud, OpenShift/Kubernetes
  • Distributed application security fabric – Granular app insights from distributed service proxies to secure web apps in real time
  • SSO / Client Authentication – SAML 2.0 authentication for back-end HTTP applications
  • Automation and programmability – REST API based solution for accelerated application delivery; extending automation from networking to developers
  • Application Analytics – Real-time telemetry from a distributed load balancing fabric that delivers millions of data points in real time

Load Balancing for VMware Horizon

NSX ALB can be configured for load balancing in VMware Horizon deployments, where you place SEs in front of Unified Access Gateways (UAG) or Connection Servers (CS) as required.

Avi Horizon High Level Architecture

For a multi-site architecture you can also configure GSLB if needed. With GSLB, access to resources is controlled with DNS queries and health checking.

Note: If you are using the Horizon Universal Broker, the cloud-based brokering service, there is no need for GSLB, because the Universal Broker can orchestrate connections from a higher level based on different policies.

Automation

With NSX Advanced Load Balancer there are two parts when we talk about automation. One part is about infrastructure automation, where the controller talks to the ecosystem like a vCenter, AWS or Azure to orchestrate the Service Engine. So, when you configure a new VIP, the controller would talk to vCenter to spin up a VM, put it in the right portgroup, connect the front and the back-end, download the policy and service engine, and starts receiveing traffic.

The second piece of automation focuses more on the operational automation which is through the REST API (the UI and CLI don’t offer all the configuration, 100% can be done via REST API). But, on top of that you can also run Ansible playbooks, Terraform templates, Go and Python SDKs, have integrations with Splunk or other tools like vRealize Automation. This is the built-in automation in the product.

Avi Networks Automation

VMworld 2020 Sessions

This year VMworld is going to be for free and virtual. Take this chance and register yourself and learn more about Avi aka NSX ALB:

  1. Making Your Private Cloud Network Run Like a Public Cloud – Part 2 [VCNC2918]
  2. Modern Apps and Containers: Networking and Security [VCNC2920]
  3. Prepare for the New Normal of Work from Anywhere [VCNC2919]

Expectations and Current Approaches

There is the general understanding and need for hybrid or multi-cloud architectures. Different people will tell different stories and give different advices. The result are different architectures and different approaches. Some people will tell you, that you can use a cloud serially, so moving from one cloud to another. Or, simultaneously, when using different services from different clouds.

My last article focused on hybrid cloud, the architecture with some services lying on the private infrastructure, while other services are hosted on a public cloud. A public cloud providers tells you, that you can buy all services from them and tries to give you a better discount than the competition (to avoid multiple clouds). Enterprises see the need for multiple (public) clouds to avoid a vendor lock-in instead of going all-in with just one of them.

VMware is about multi-cloud and workload mobility, with the vision, that their VCF stack is running everywhere in the future. Now, some people would now say that this is also a vendor lock-in. Depending on your strategy and technology choices and preferences (e.g. databases, AI/ML services, virtual desktops), you have to decide somewhen which (cloud) vendor, approach and operation model is the right one for you.

It may not true for every large environment, but if you go for multiple clouds, multiple technologies, management and security consoles, architecture and so on, you’ll spend a lot of time and money on engineering and keeping your environment “integrated” and functional.

VMware offers you choice. The choice to run your workloads today and tomorrow wherever you want.

If you have the same vision and strategy like VMware, then you are looking for solutions which run in or on top of every cloud. Because of that it’s very important to understand the different between multiple clouds and multi-cloud.

In this case, NSX ALB brings you multi-cloud load balancing and auto scaling features for any cloud and for multi-cloud enabled applications and services.

Don’t forget: Some people are also saying,  that multi-cloud is not needed and doesn’t exist in reality. Nobody is saying multi-cloud is a piece of cake, but VMware can definitely help you to abstract this complexity. And part of this abstraction can be handled with vRealize Automation for example, which can act as a cloud broker to deploy your application and services.

 

Azure VMware Solution

Azure VMware Solution

Update May 2020:

On May 4th Microsoft announced the preview (and the “next evolution”) of Azure VMware Solution which is now a first-party offering service designed, built and supported my Microsoft and endorsed by VMware. This is an entirely new service entirely delivered and supported by Microsoft and does not replace the current AVS solution/service by CloudSimple at this time. This is truly just a Microsoft technology offering and has also nothing to do with a Virtustream branded Azure VMware Solution offering. Short: A way cleaner offering and service with a contract only between Microsoft and VMware.

— Original Text below —

Since Dell Technologies World 2019 it’s clear: VMware and Microsoft are not frenemies anymore!

Dell Technologies and Microsoft announced an expanded partnership which should help customers and provide them more choice and flexibility for their future digital workspace projects or cloud integrations.

One result and announcement of this new partnership is the still pretty new offering called “Azure VMware Solution” (AVS). Other people and websites may also call it “Azure VMware Solution by Virtustream” or “Azure VMware Solution by CloudSimple”.

AVS is a Microsoft first-party offering. Meaning, that it’s sold and supported by Microsoft, NOT VMware. This is one very important difference if you compare it with VMC on AWS. The operation, development and delivery are done by a VMware Cloud Verified and Metal-as-a-Service VCPP (VMware Cloud Provider Program) partner; CloudSimple or Virtustream (subsidiary of Dell Technologies). AVS is fully supported and verified by VMware.

VMware Metal-as-a-Service Authorized partners Virtustream and CloudSimple run the latest VMware software-defined data center technology, ensuring customers enjoy the same benefits of a consistent infrastructure and consistent operations in the cloud as they achieve in their own physical data center, while allowing customers to also access the capabilities of Microsoft Azure.

So, why would someone like Microsoft run VMware’s Cloud Foundation (VCF) stack on Azure? The answer is quite simple. VMware has over 500’000 customers and an estimated number of 70mio VMs which are mostly running on-premises. Microsoft’s doesn’t care if virtual machines (VMs) are running on vSphere, they care about Azure and the consumption in the end. AVS is just another form of Azure, Microsoft says. I would say it’s very unlikely that a customer moves on to Azure native once they are onboarded via Azure VMware Solution.

Microsoft would like to see some of the 70mio running on their platform, no matter if it’s VCF on top of their Azure servers. Customers should get the option to move to the Azure cloud, using Azure native services (e.g. Azure NetApp Files, Azure databases etc.), but give them the choice and flexibility to use their existing technology stack, ecosystem and tools (e.g. automation or operation) they are familiar with – the whole or some part of the VCF coupled with products from the vRealize Suite. Plus, other VMware 3rd party integrations they might have for data protection or backup. This is one unique specialty – Microsoft says – that there is no restricted functionality as you may experience in other VMware clouds.

Azure VMware Solutions Components

From VMware’s perspective most of our customers are already Microsoft customers as well. In addition to that VMware’s vision is to provide the freedom of choice and flexibility, same like Microsoft, but it one small difference: to be cloud and infrastructure agnostic. This vision says that VMware doesn’t care if you run your workloads on-prem, on AWS, Azure or GCP (or even at a VCPP partner’s cloud) as long it’s running on the VCF stack. Cloud is not a choice or destination anymore, it has become an operation model.

And to keep it an operation model without having a new silo and the vendor lock-in, it makes totally sense to use VMware’s VCF on top of AWS, Azure, GCP, Oracle, Alibaba or any other VCPP partners. This ensures that customers have the choice and flexibility they are looking for, coupled with the new and maybe still surprising “new” or “special” public cloud. If your vision is also about workload mobility on any cloud, then VMware is the right choice and partner!

Use Cases

What are the reasons to move to Azure and use Azure VMware Solution?

If you don’t want to scale up or scale out your own infrastructure and would like to get additional capacity almost instantly, then speed is definitely one reason. Microsoft can spin up a new AVS SDDC under 60min, which is impressive. How is this possible? With automation! This proves that VMware Cloud Foundation is the new data center operating system of the future and that automation is a key design requirement. If you would like to experience nearly the same speed and work with the same principles as public cloud provider do, then VCF is the way to go.

The rest of the use cases or reasons are in general the same if we talk about cloud. If it’s not only speed, then agility, (burstable) capacity, expansion in a new geography, DRaaS or for app modernization reasons using cloud native services.

Microsoft Licenses

What I have learned from this MS Ignite recording, is, that you can bring your existing MS licenses to AVS and that you don’t have to buy them AGAIN. In any other cloud this is not the case.

This information can be found here as well:

Beginning October 1, 2019, on-premises licenses purchased without Software Assurance and mobility rights cannot be deployed with dedicated hosted cloud services offered by the following public cloud providers: Microsoft, Alibaba, Amazon (including VMware Cloud on AWS), and Google. They will be referred to as “Listed Providers”.

Regions

If you check the Azure documentation, you’ll see that AVS is only available in US East and West Azure regions, but should be available in Western Europe “in the near future”. In the YouTube video above Microsoft was showing this slide which shows their global rollout strategy and the planned availability for Q2 2020:

Azure VMware Solutions Regions 2020

According to the Azure regions website Azure VMware Solution is available at the following locations and countries in Europe:

Azure VMware Solutions by Azure RegionSo, North Europe (UK) is expected for H2 2020 and AVS is already available in the West Europe Azure region. Since no information available about the Swiss regions, even the slide from the MS Ignite recording may suggest the availability until May 2020, it’s very unlikely that AVS will be available in Zurich or Geneva before 2021.

Azure VMware Solution Components

You need at least three hosts to get started with the AVS service and you can scale up to 16 hosts per cluster with a SLA of 99.9%. More information about the available node specifications for your region can be found here. At the moment CloudSimple offers the following host types:

  • CS28 node: CPU:2x 2.2 GHz, total 28 cores, 48 HT. RAM: 256 GB. Storage: 1600 GB NVMe cache, 5760 GB data (All-Flash). Network: 4x25Gbe NIC
  • CS36 node: CPU 2x 2.3 GHz, total 36 cores, 72 HT. RAM: 512 GB. Storage: 3200 GB NVMe cache 11520 GB data (All-Flash). Network: 4x25Gbe NIC
  • CS36m node (only option for West Europe): CPU 2x 2.3 GHz, total 36 cores, 72 HT. RAM: 576 GB. Storage: 3200 GB NVMe cache 13360 GB data (All-Flash). Network: 4x25Gbe NIC

I think it’s clear that the used hypervisor is vSphere and that it’s maintained by Microsoft and not by VMware. There is no host-level access, but Microsoft gives you the possibility of a special “just in time” privileges access (root access) feature, which allows to install necessary software bits you might need – for example for 3rd party software integrations.

The storage infrastructure is based on vSAN with an all-flash persistent storage and a NVMe cache storage. More capacity can be made available by adding additional nodes or use Azure offerings which can be added to VMs directly.

Networking and security are based on NSX-T which fully supports micro segmentation.

To offer choice, Microsoft gives you the option to manage and see your AVS VMware infrastructure via vCenter or Azure Resource Manager (ARM). The ARM integration will allow you to create, start, stop and delete virtual machines and is not meant to replace existing VMware tools.

Microsoft support is your single point of contact and CloudSimple contacts VMware if needed.

Connectivity Options

CloudSimple provides the following connectivity options to connect to your AVS region network:

Depending on the connectivity option you have different ways to bring your VMs to your AVS private cloud:

How do I get started?

You have to contact your Microsoft account manager or business development manager if would like to know more. But VMware account representatives are also available to support you. If you want to learn more, check https://aka.ms/startavs.

Can I burn my existing Azure Credits?

Yes. Customers with Azure credits can use them through Azure VMware Solution.

VMware’s Tanzu Kubernetes Grid

VMware’s Tanzu Kubernetes Grid

Since the announcement of Tanzu and Project Pacific at VMworld US 2019 a lot happened and people want to know more what VMware is doing with Kubernetes. This article is a summary about the past announcements in the cloud native space. As you already may know at this point, when we talk about Kubernetes, VMware made very important acquisitions regarding this open-source project.

VMware Kubernetes Acquisitions

It all started with the acquisition of Heptio, a leader in the open Kubernetes ecosystem. With two of the creators of Kubernetes (K8s), namely Joe Beda and Craig McLuckie, Heptio should help to drive the cloud native technologies within VMware forward and help customers and the open source community to accelerate the enterprise adoption of K8s on-premises and in multi-cloud environments.

The second important milestone was in May 2019, where the intent to acquire Bitnami, a leader in application packaging solutions for Kubernetes environments, has been made public. At VMworld US 2019 VMware announced Project Galleon to bring Bitnami capabilities to the enterprise to offer customized application stacks to their developers.

One week before VMworld US 2019 the third milestone has been communicated, the agreement to acquire Pivotal. The solutions from Pivotal have helped customers learn how to adopt modern techniques to build and run software and they are the provider of the most popular developer framework for Java, Spring and Spring Boot.

On the 26th August 2019, VMware gave those strategic acquisitions the name VMware Tanzu. Tanzu should help customers to BUILD modern applications, RUN Kubernetes consistently in any cloud and MANAGE all Kubernetes environments from a single point of control (single console).

VMware Tanzu

Tanzu Mission Control (Tanzu MC) is the cornerstone of the Tanzu portfolio and should help to relieve the problems we have or going to have with a lof of Kubernetes clusters (fragmentation) within organizations. Multiple teams in the same company are creating and deploying applications on their own K8s clusters – on-premises or in any cloud (e.g. AWS, Azure or GCP). There are many valid reasons why different teams choose different clouds for different applications, but is causing fragmentation and management overhead because you are faced with different management consoles and silo’d infrastructures. And what about visibility into app/cluster health, cost, security requirements, IAM, networking policies and so on? Tanzu MC let customers manage all their K8s clusters across vSphere, VMware PKS, public cloud, managed services or even DIY – from a single console.

Tanzu Mission Control

It lets you provision K8s clusters in any environment and configure policies which establish guardrails. Those guardrails are configured by IT operations and they will apply policies for access, security, backup or quotas.

Tanzu Mission Control

As you can see, Mission Control has a lot of capabilities. If you look at the last two images you can see that you not only can create clusters directly from Tanzu MC, but also have the ability to attach existing K8s clusters. This can be done by installing an agent in the remote K8s cluster, which then provides a secure connection back to Tanzu MC.

We focused on the BUILD and MANAGE layers now. Let’s take a look at the RUN layer which should help us to run Kubernetes consistently across clouds. Without consistency across cloud environments (this includes on-prem) enterprises will struggle to manage their hundred or even thousands of modern apps. It’s just getting too complex.

VMware’s goal in general is to abstract complexity and to make your life easier and for this case VMware has announced the so-called Tanzu Kubernetes Grid (TKG) to provide us a common Kubernetes distribution across all the different environments.

Tanzu Kubernetes Grid

In my understanding TKG means VMware’s Kubernetes distribution, will include Project Pacific as soon as it’s GA and is based on three principles:

  • Open Source Kubernetes – tested and secured
  • Cluster Lifecycle management – fully integrated
  • 24×7 support

Meaning, that TKG is based on open source technologies, packaged for enterprises and supported by VMware’s Global Support Services (GSS). Based on these facts you could say, that today your Kubernetes journey with VMware starts with VMware PKS. PKS is the way VMware deliver the principles of Tanzu today – across vSphere, VCF, VMC on AWS, public clouds and edge.

Project Pacific

Project Pacific, which has been announced at VMworld US 2019 as well, is a complement to VMware PKS and will be available in a future release. If you are not familiar with Pacific yet, then read the introduction of Project Pacific. Otherwise, it’s sufficient to say, that Project Pacific means the re-architecture of vSphere to natively integrate Kubernetes. There is no nesting or any kind of it and it’s not Kubernetes in vSphere. It’s more like vSphere on top of Kubernetes since the idea of this project is to use Kubernetes to steer vSphere.

Project Pacific

Pacific will embed Kubernetes into the control plane of vSphere and converge VMs and containers on the same underlying platform. This will give the IT operators the possibility to see and manage Kubernetes from the vSphere client and provide developers the interfaces and tools they are already familiar with.

Project Pacific Console

If you are interested in the Project Pacific Beta Program, you’ll find all information here.

I would have access to download the vSphere build which includes Project Pacific, but I haven’t got time at the moment and my home lab is also not ready yet. We hear customers asking about the requirements for Pacific. If you watch all the different recordings from the VMworld sessions about Project Pacific and the Supervisor Cluster, then we could predict, that only NSX-T is a prerequisite to deploy and enable Project Pacific. This slide shows why NSX-T is part of Pacific:

Project Pacific Supervisor ClusterFrom this slide (from session HBI1452BE) we learn that a load balancer built on NSX Edge is sitting in front of the three K8s Control Plane VMs and that you’ll find a Distributed Load Balancer spanned across all hosts to enable the pod-to-pod or east-west communication.

Nobody of the speakers ever mentioned vSAN as a requirement and I also doubt that vSAN is going to be a prerequisite for Pacific.

You may ask yourself now which Kubernetes version will be shipped with ESXi and how you upgrade your K8s distribution? And what about if this setup with Pacific is too “static” for you? Well, for the Supervisor Clusters VMware releases patches with vSphere and you apply them with the known tools like VUM. For your own built K8s clusters, or if you need to deploy Guest Clusters, then the upgrades are easy as well. You just have to download the new distribution and specify the new version/distribution in the (Guest Cluster Manager) YAML file.

Conclusion

Rumos say that Pacific will be shipped with the upcoming vSphere 7.0 release, which even should include NSX-T 3.0. For now we don’t know when Pacific will be shipped with vSphere and if it really will be included with the next major version. I would be impressed if that would be the case, because you need a stable hypervisor version, then a new NSX-T version is also coming into play and in the end Pacific relies on these stable components. Our experience has shown that the first release normally is never perfect and stable and that we need to wait for the next cycle or quarter. With that in mind I would say that Pacific could be GA in Q3 2020 or Q4 2020. And beside that the beta program for Project Pacific just has started!

Nevertheless I think that Pacific and the whole Kubernetes Grid from VMware will help customers to run their (modern) apps on any Kubernetes infrastructure. We just need to be aware that there are some limitations when K8s is embedded in the hypervisor, but for these use cases Guest Clusters could be deployed anyway.

In my opinion Tanzu and Pacific alone don’t make “the” big difference. It’s getting more interesting if you talk about multi-cloud management with vRA 8.0 (or vRA Cloud), use Tanzu MC for the management of all your K8s clusters, networking with NSX-T (and NSX Cloud), create a container host with a container image (via vRA’s Service Broker) for AI- and ML-based workloads and provide the GPU over the network with Bitfusion.

Bitfusion Architecture

Looking forward to such conversations! 😀

VCAP7-DTM Design Exam, Part 10

In part 10 of my VCAP7-DTM Design exam series we take a look at the Horizon 7 Enterprise Reference Architecture.

To be honest, I didn’t study that much the last two weeks but I checked a few documents about App Volumes, Mirage, ThinApp and User Environment Manager.

This time I would like to summarize what I have learned from the reference architecture and the VMworld 2018 session called Architecting Horizon 7 Enterprise: The Official Reference Architecture (WIN3451BUR).

I only focus on the component design part since I already covered topics like use cases, business drivers, design methodology etc.

Horizon 7

A successful deployment depends on good planning and a very good understanding of the platform. The core elements include Connection Server, Composer, Horizon Agent and Horizon Client. Part 4 to part 9 cover the Horizon 7 component design and also provide more information on the following components.

Horizon 7 Logical Architecture

Identity Manager

VMware Identity Manager (VIDM) can be implemented on-premises or in the cloud, a SaaS-based implementation. If you decide to go with the SaaS implementation, a VIDM connector needs to be installed on-prem to synchronize accounts from Active Directory to the VIDM service in the cloud.

If cloud is no option for you, you still have the possibility for the on-prem deployment and use the Linux-based virtual appliance. There is also a Windows-based installer available which is included in the VMware Enterprise Systems Connector. VMware’s reference architecture is based on the Linux appliance.

VMware Identity Manager Architecture

Syncing resources such as Active Directory and Horizon 7 and can be done either by using a separate VMware Identity Manager Connector or by using the built-in connector of an on-premises VMware Identity Manager VM. The separate connector can run inside the LAN in outbound-only connection mode, meaning the connector receives no incoming connections from the DMZ.

VIDM comes with an embedded PostgreSQL database, but it’s recommended to use an external database server for production deployments.

For high availability, based on your requirements, at least two VIDM appliances should be deployed behind a load balancer. After you have deployed your first appliance, you simply clone it and assign a new hostname and a new IP address.

App Volumes

As you still may know from part 8, App Volumes has two functions. The first is the delivery of applications for VDI and RDSH. The second is the provision of writable volumes to capture user-installed applications and the user profile.

app volumes architecture

For high availability, always use at least two App Volumes Managers which are load-balanced.

AppStacks are very read intensive, hence, you should place AppStacks on storage that is optimized for read operations. Writable volumes should be placed on storage for random IOPS (50/50). There reference architecture uses vSAN to provide a single highly available datastore.

For the SQL database it is recommended using an AlwaysOn Availability Group.

User Environment Manager

When User Environment Manager design decisions need to be made, you have to think about user profiles (mandatory, roaming, local) and folder redirection. As already described in part 9, VMware recommendation is to use mandatory profiles and folder redirection. Use appendix B if you need help configuring the mandatory profile.

vmware user environment manager

The first key design consideration is using DFS-R to provide high availability for the configuration and user shares. Note: Connect the management console only to the hub member when making changes. DFS-R will replicated those changes to the spoke members.

The second consideration one is using GPO loopback processing.

Unified Access Gateway

In part 6 I mentioned that a UAG is typically deployed within the DMZ.

VMware Unified Access Gateway

UAG appliances are deployed in front of the Horizon 7 Connection Servers and sit behind a load balancer. The Unified Access Gateway also runs the Content Gateway as part the AirWatch (WorkspaceONE UEM) service.

You have two sizing options during the appliance deployment:

  • Standard (2 vCPU, 4GB RAM, 2’000 Horizon server connections, 10’000 AirWatch service connections)
  • Large (4 vCPU, 16GB RAM, 2’000 Horizon server connections, 50’000 AirWatch service connections)

As you can see, the big difference here are the estimated AirWatch service connections per appliance. In production you would deploy dedicated UAG appliances for each service. Example:

  • 2 standard size UAGs appliances for 2’000 Horizon 7 sessions (n+1)
  • 3 large size UAG appliances for 50’000 devices using Content Gateway and per-App Tunnel which gives us a total of 100’000 sessions. The third appliance is for high availability (n+1)

vSphere and Physical Environment

The software-defined data center (SDDC) is the foundation that runs all infrastructure servers and components. The products and the licensing for the foundation are outside of the Horizon 7 product (except vSAN), but are required to deliver a complete solution.

And in my opinion this is what makes the whole solution so brilliant. Even I work for VMware, I would never say from the beginning that Horizon is better than XA/XD. This was also the case when I worked as a consultant for Citrix before I joined VMware in May 2018.
It depends on the requirements and use cases which need to be satisfied. That are the most important things if you choose a vendor or a specific technology. Our goal is to make the customer happy! 🙂

But I would say that VMware Horizon including WorkspaceONE is very hard to beat if you use the complete stack! But that’s another topic.

The vSphere infrastructure in the reference architecture includes vSAN and NSX. In part 5 I covered the basics of vSAN, but I think I maybe need to write a short overview about NSX and how you can use it with Horizon.

vSAN provides a hyper-converged storage optimized for virtual machines without the need for an external SAN or NAS. This means that the physical server not only provides the compute and memory resources, but also storage in a modular fashion. You can use vSAN for the management and resource block  and follow a hybrid approach for the management resources and use all-flash vSAN for the Horizon resources.

VMware vSAN

I will not cover the vSphere design, but it’s important to understand that all components are operating redundantly and that you have enough physical resources to meet the requirements.

vSphere Networking

A general recommendation is to use at least 10 GbE connections, to separate each traffic (mgmt, VM traffic, vSAN, vMotion) and make sure that each of them has sufficient bandwidth.

NSX for vSphere

NSX provides several network-based services and performs several security functions within a Horizon 7 implementation:

  • Protects VDI infrastructure
  • Protects desktop pool VM communication with applications
  • Provides user-based access control (user-level identity-based micro-segmentation)

VMware NSX for vSphere

If you want to use NSX you have to think about a NSX infrastructure design as the NSX platform adds new components (e.g. NSX manager) and new possibilities (distributed firewall and identity firewall).

The most important design consideration for Horizon 7 is the concept of micro-segmentation. In the case of Horizon 7, NSX can block desktop-to-desktop communications, which are normally not needed or recommended. Each VM can now be its own perimeter and this desktop isolation prevents threats from spreading:

NSX isolation

The Horizon 7 reference architecture of probably the best document to prepare yourself for the VCAP7-DTM exam. What do the current VCAP7-DTM certified  people say? What else needs to be covered? Jump to part 11

VCAP7-DTM Design Exam, Part 6

This is the sixth part of my VCAP7-DTM Design exam series. In part 5 I covered the creation of a physical design for horizon storage. This time we take a look at section 5 of the blueprint, the creation of a physical network design for Horizon:

Section 5 – Create a Physical Design for Horizon Networking
Objective 5.1 – Plan and Design Network Requirements for Horizon solutions (including Mirage and Workspace One)
Objective 5.2 – Design Network and Security Components Based on Capacity and Availability Requirements
Objective 5.3 – Evaluate GPO and Display Protocol Tuning Options Based on Bandwidth and Connection Limits

Networking is also a very important and exciting when creating a Horizon architecture and a lot of questions are coming up when I think about Horizon and network access and devices:

  • How does the ISP infrastructure look like?
  • Do we have redundant internet uplinks?
  • Bandwidth in the data center?
  • Firewalls?
  • Remote connections?
  • How is the connection between Horizon client and agent?
  • ESXi host network interfaces?
  • Do we have mobile workers using WLAN?

I once had a customer who had a really nice and modern data center infrastructure, but their firewalls didn’t provide enough throughput. Make your homework and know how the routing and switching looks like and check every component’s limit.

Beside our VDI traffic, what about management, vMotion and vSAN traffic? Do we have enough network interfaces and bandwidth? If you think about management traffic, then 1Gbit interfaces are normally sufficient. But vMotion and vSAN traffic should have redundant 10Gbit connections and be on different subnets/VLANs.

Overview of the Network Architecture

In most network architectures two firewalls exist to create the DMZ.

The Unified Access Gateway (UAG) appliances are placed in the DMZ. UAG can perform authentication or pass a connection to the Connection Server for AD authentication.
Notauthenticated sessions are dropped at the Unified Access Gateway appliance and only authenticated sessions are allowed to connect to the internal resources.

UAG appliances in the DMZ communicate with the Connection Server instances inside the corporate firewalls and ensure that only the desired remote apps and desktop sessions can enter the corporate data center on behalf of this strongly authenticated user.

Inside the corporate firewall you install and configure at least two Connection Server instances. Their configuration data is stored in an embedded LDAP directory (AD LDS) and is replicated among all members of the group.

Firewall Ports

On March 22, 2016, an updated network ports diagram has been posted by VMware:

Horizon 7 Network Ports Diagram

On Tech Zone this diagram and all key firewall considerations are available for Horizon 7: https://techzone.vmware.com/resource/network-ports-vmware-horizon-7

Network Bandwidth Considerations

The used session bandwidth between the Horizon client and agent depends highly on the session configuration. For display traffic, many elements can affect network bandwidth, such as the used protocol, monitor resolution, frames per second, graphically intense applications or videos, image and video quality settings.

Because the effects of each configuration can vary widely, it’s recommended to monitor the session bandwidth consumption as part of a pilot. Try to figure out the bandwidth requirements for each use case.

Display Protocol

I would say that Blast Extreme is the way to go, because it has been optimized for mobile devices and can intelligently switch between UDP and TCP (Adaptive Transport). PCoIP has been developed by Teradici, but Blast is VMware’s own creation and that’s why I think that Blast will be “the future” and that RDP still can be used as fallback for some special scenarios.

Display Protocol Tuning Options

I will not cover this topic and explain you how you can configure the maximum bandwidth for PCoIP via GPO. There are several options to decrease and increase the used session bandwidth:

Configuring PCoIP session variables
VMware Blast Policy Settings

WAN Consideration

Nowadays, every client device is connected with 1Gbps. LAN connections and the user experience are most of the time perfect. How is it with WAN connections where you will have latencies that could be between 50 and 200ms? Do you apply Quality of Services (Qos) policies to prioritize Horizon traffic?

WAN optimization is one of the keywords when talking about WAN connections and is valuable for TCP-based protocols which require many handshakes between client and server, such as RDP.
PCoIP is  UDP-based and this was the reason why everyone in the past said, that you should prefer this protocol for connections with higher latencies and then no WAN optimization or acceleration would be needed.

Then inside the corporate network you would use RDP because your network is stable or did you leave this choice to the user?

With Blast Extreme, Adaptive Transport will automatically detect higher latencies and automatically switches between TCP and UDP if needed. Higher latencies could also occur with mobile devices working of WiFi networks.

In my opinion there are almost no reasons anymore to use anything else than Blast because it’s also more network efficient than PCoIP.

pcoip blast extreme comparison

Conclusion

Use separate networks for vSphere management, VM connectivity, vMotion and vSAN traffic. Make sure you have redundancy across different physical adapters (NIC, PCI slot) and devices (switches, router, firewall). Consider the use of a vSphere Distributed Switch (vDS) to reduce management overhead and provide a richer feature set. Maybe NSX could be interesting for micro segmentation.

Load balancing is a very important component of a Horizon architecture. The primary purpose of load balancing is to optimize performance by evenly distributing client sessions across all available Connection Server instances. The same is valid for UAG appliances, Identity Manager or App Volumes Manager. NSX comes with a virtual load balancer, but F5 and NetScaler are also fine.

Depending on your customer’s requirements and needs, the network design is another key part to remove single point of failures.

In part 7 we will figure out how we have to design Horizon desktops and pools.