When VMware Cloud on AWS was announced in 2017, it was met with a mix of optimism and reservations, because it was unknown if it was going to succeed as a solution. However, the last year has shown us that it has been wildly successful with huge projected growth for the next year.
To understand why it has been so successful and what some people were worried about, you have to understand the history. First, we will go back a decade to 2008, when the VMware vCloud initiatives were first created. These initiatives focused on 3 things:
- Private clouds hosted internally
- Private clouds hosted by 3rd parties
- Virtual Desktop Infrastructure (VDI)
Now, let’s drill into what the evolution of each initiative has morphed in to over the last decade.
Private clouds hosted internally
11 years ago, VMware starting really moving away from the ESXi hypervisor as the “go to” product offering. The core ESXi product was being given away for free and the solution value was moving more towards high availability, manageability, and scalability. The Virtual Datacenter OS (VDC-OS) was created to address high availability, fault-tolerance, basic network virtualization, and data recovery.
This evolved on a few different fronts:
- Deployment of applications were becoming more orchestrated and self-contained with VMware Orchestrator and vApps. This evolved over time to incorporate vCloud Director for service providers and vRealize Automation for internal IT deployment automation and management.
- Management and operational insight were increasing with the use of vRealize Operations (initially it was named vCOPs for vCloud Operations). vRealize Configuration Manager was also added to provide insight into risk, governance, and compliance.
- Storage virtualization removed the need for many 3rd party storage vendors, dedicated storage networks, and SANs. This evolved into what is currently VSAN.
- Network virtualization was initially provided a long time ago with virtual switching, but it evolved to incorporate virtual routers, load balancers, firewalls, VPN services, and security services. VMware created vCloud Networking and Security (VCNS) which had many features, but this was superseded by NSX when Nicira was acquired in 2012.
The combination of these technologies became known the Software Defined Datacenter, or SDDC. This is the core building block of VMware’s solution offerings.
Private clouds hosted by 3rd parties
When service providers first started offering infrastructure as a service, they would build custom environments based on their internal build and architectural practices. This would invariably lead to issues with version changes, workload migrations, security, and scalability. This was then addressed with the vCloud initiatives for data centre’s. Standard building blocks that could be used to provide resources, manage, and scale them. vCloud service provider partners would build the environment based on the standard, then provide a front-end to customers based on vCloud Director so they could utilize the self-service architecture.
To make life even easier for customers, VMware decided that they were going to create an easy on-board ramp to get to the cloud. This was called vCloud Hybrid Service (also known with the unfortunate moniker vCHS or vCheese). Later, the service was renamed as vCloud Air. The service was run internally by a division with VMware using the SDDC as the underlying infrastructure. Competition as a public cloud provider was very difficult with fewer regions and availability zones than Amazon, Azure, or GCP. The benefit of using vCloud Air was the native integration with vSphere using the vCloud Connector. With the connector, you could move workloads from one private cloud to vCloud Air or a VMware vCloud provider.
As of May 2017, VMware has divested of the vCloud Air division. It was acquired by a company named OVH and has since been renamed OVHcloud.
Virtual Desktop Infrastructure (VDI)
VMware VDM (Virtual Desktop Manager) was created in 2005 and presented at VMworld as the first broker-based desktop connection solution. Before then, you would have to use Microsoft RDSH or Citrix to connect to a session-based desktop on a server. After several iterations, VDM was renamed to View at version 3.0 in 2008. In 2014, it was again renamed to VMware Horizon. The cloud pod architecture was then created to allow easy, pod-based deployments using the SDDC as the base building block. Now you can deploy virtual desktops on premises, in a VMware managed public cloud infrastructure, or on the Microsoft Azure public cloud.
VMware Cloud Foundation
As more and more features and services get added to the core SDDC stack, it becomes difficult to manage dependencies, patches, and operational workflows for management. This was addressed with VMware Cloud Foundation. VCF provides a method of securing, simplifying, and standardizing infrastructures across on-prem, 3rd party hosted, and public cloud infrastructures. It applies lifecycle management to the entire architecture and compatibility to ensure that workloads can be migrated and interop seamlessly between different environments.
What is VMware Cloud on AWS?
VMC on AWS is a managed and instantly scalable implementation of VMware Cloud Foundation. It runs on baremetal in Amazon Web Services and connects with both your on-prem vSphere environment and native AWS services. This allows you migrate workloads back and forth between on-prem and VMC on AWS, as well as securely access all workloads running natively in AWS, with no ingress / egress fees. You have the ability to scale nodes as required on a per-minute basis, based on the workload demand. You can run VM workloads, or containers in the cloud, either on VMC on AWS, native AWS, or a mix of the two.
Here are some use cases and the benefits you get from each one.
It does not make sense to run certain applications in AWS, because they do not gain all the benefits, and may end up costing more to run there. The ideal type of application to run in AWS is cloud native, stateless, and has dynamically elasticity. Traditional legacy applications are not designed to scale out and thus have monolithic components that scale up instead. VMC on AWS provides a solution for gaining some of the benefits of the cloud, without re-architecting applications. This provides an in-between state for the modernization process.
During the period of time that an organization is modernizing their application architecture, there may be several hardware generations that they transition between. Infrastructure refreshes have to be planned and orchestrated to minimize impact and provide maximum compatibility to support mixed generation environments. VMC on AWS has hardware refreshes and infrastructure upgrades integrated as part of the managed service. This reduced staff costs, maintenance windows, and the capex of hardware lifecycle purchases.
Data Center Extension
As workloads get added to an environment, a decision needs to be made where they will run. Will the existing infrastructure support the additional load? Will there be dedicated equipment purchased for it? Does it require isolation from other workloads? Are there internal or external SLAs that will be applied to it? How long will the resources be needed? Long term or short term?
The answers to these questions end up falling onto a few realities:
- It’s always cheaper to use what you already have, and until you no longer have the resources, then that’s the most common place for a new workload to go.
- To ensure availability during a failure, the infrastructure component (compute, storage, network) utilization cannot exceed a certain safe threshold.
- Buying new equipment for a workload is a long-term strategy. It is not cost effective to do so otherwise.
- Deploying and managing new equipment adds more work for existing staff. Depending on the current utilization, additional staff may also have to be hired.
- If a 3rd party service provider is used, you will need to know the delineation of where their service ends and internal responsibility begins.
- There needs to be a way to move workloads and data between existing and new resources.
- Short term resources (ie: the cloud) are much more expensive than long-term (on-prem) resources, after a certain amount of use.
- External availability and performance are a function of the reliability and redundancy of the upstream network connectivity.
This is where VMC on AWS fits in.
- It provides a large block of resources that are highly available, with very good SLAs.
- It has cloud-centric elasticity to grow or contract as needed for short term workloads
- It has higher operational costs than on-prem, but the cost is inclusive of a managed infrastructure and the ability to run at much higher consolidation thresholds.
- It is a good fit if some companies are merging some applications need to be ingested and owned by another team. Migrate the workloads into VMC on AWS and decommission the old hardware. The risk of managing old and unknown infrastructure is averted.
- If developers create an application in house, that they want to test at scale without refactoring it, then VMC on AWS can scale on demand to any size. If blueprints are being used, then where the resources get provisioned can be modified very easily.
- Applications could also be validated and deployed at scale before dedicated hardware is purchased, as a means of reducing the time to market.
Disaster recovery reduces risk but adds cost and complexity. First, let’s differentiate backup from DR. Backup provides you with a point in time copy of your data, whereas DR is a process that uses copies of your data in a workflow to make applications and services available after a service impacting event. DR may entail hunting down tapes in a storage facility, buying a tape drive, network equipment, servers and some storage, then installing operating systems and restoring data from those tapes. Or it may involve a secondary site that data is being replicated to and a recovery workflow engine that orchestrates the fail-over and fail-back process, like with VMware Site Recovery Manager.
With backup and DR, there are a few things that are very important.
- The rule of 3
- 3 copies of the most important data – Two isn’t enough if it’s important.
- 2 different formats – Storage based snapshots and immutable archives.
- 1 off-site backup – Another site, an offsite secure vault, the cloud.
- The RPO/RTO and retention requirements of your data
If you want to ensure that your recovery time objective (RTO) is as low as possible, then you need to have another site with resources available to stand up your infrastructure. This can be very costly to maintain if the requirements are to have standby resources that are not actually being utilized. The TCO for the entire environment may double. You can reduce the TCO by having a production site and a dev/DR site. The production site would run the primary workloads, replicating them to the secondary site. The secondary site would run workloads that do not have the primary workload SLA. Thus, if there is a failure, then the secondary site would shut down its servers to accommodate the workload of the primary site that has failed over.
VMC on AWS can act as a DR site for either standby resources, or a full secondary site. The difference is in the number of nodes that are allocated. It is possible to have just a single host, 3 hosts or 4+ hosts. You can reduce the cost of having a dedicated site by making use of DR as a Service, with VMC on AWS.
If VMC on AWS is your production site, you can also use SRM to have a multi-region DR strategy, to another VMC on AWS instance, or to your on-premises environment. It is very easy to configure and could update, augment, or replace your existing DR strategy.
Virtual Desktop Infrastructure
Because of the design of the Cloud Pod Architecture for Horizon, it is easy to utilize VMC on AWS for your VDI environment. It could be your primary environment, or your DR environment, or a mix of the two. If you have a DR event and you need to scale your VDI environment to support thousands of more users, you can easily do that by scaling the number of hosts that are provisioned, then reducing them when the event has been resolved. No need to pre-allocate the hosts and not have them utilized.
Scalar is an AWS Advanced Consulting Partner and a VMware Premier Solution Provider Partner. We have expertise in helping clients assess, design, cost model, and deploy architectures that can support multi-region hybrid cloud environments. So, if you need any help, please do drop us a line!