06 Oct Hyperconvergence and the Public Cloud
It’s important to understand the ways by which you can leverage cloud services as a part of your hyperconverged infrastructure solutions. It’s also important to understand the concept of “private cloud” and how that fits with hyperconvergence. Can a hyperconverged solution deliver some of the things that cloud can give you?
Why Is Cloud so Desirable?
We will cover what defines cloud a little later. Before getting into the various definitions, though, let’s discuss the traits inherent in cloud systems which make them a popular and desirable choice for service deployment.
The Economic Model
Everything eventually comes down to money. Business decision makers are constantly on the lookout for ways to reduce costs while also boosting efficiency and outcomes. This is often a seemingly impossible task described as “doing more with less.” IT was supposed to be an enabler, but for many companies, it has become a money pit — an expense center to be minimized. Obviously, when leveraged properly, IT can be an incredible enabling function, but even in these cases, no one wants to spend more than they have to.
Economic Drivers for Technology Acquisition
There are three primary metrics that decision makers seek to satisfy when embarking on technology investments:
- Return on investment (ROI) — Any time you make any kind of investment, you expect to see some kind of result from that investment. For example, if you buy stocks, you expect to see the stock price rise over time and eventually pay you back more than your initial cash investment. In the data center, an ROI can come in different forms. You may be able to directly save money in another area by making a technology investment. Or, you may be able to save time on a process. In a business, saving time equates to saving money. The goal with ROI is to get back more than you put in. At the very least, your ROI should result in not moving backward.
- Initial cost — If you can’t afford the initial cash outlay for something, even the highest ROI and lowest total cost of ownership (TCO) won’t matter. That’s one of the reasons that the initial cost for solutions remains of primary importance for many organizations.
- Total cost of ownership (TCO) — Everything you buy in the data center requires additional funding to maintain. This is beyond the initial cost of the solution. For example, for hardware and software alike, you generally have to pay annual maintenance fees. You may need to hire staff to maintain the solution. Additional staff costs money. You may incur installation charges or other fees. Everything that it takes to support a solution, including its initial cost, is considered the total cost that it takes to keep it — the total cost of ownership.
In addition, there are two kinds of expenses that you need to keep in mind when you’re considering data center economics:
- Operational expenditures (OpEx) — OpEx generally aligns with the TCO of a solution minus the initial costs. These are the expenses that are incurred when administering and maintaining what you’ve already purchased.
- Capital expenditures (CapEx) — These generally align pretty closely with the initial cost of a solution and are often considered the one-time costs associated with a purchase.
When you buy your own data center hardware and software, you incur pretty significant CapEx. This initial cash outlay necessary to procure a solution can be pretty high and can result in the need to cut corners or even delay upgrades if there is not enough cash available.
When you decide to start consuming resources from the public cloud, there is no initial cash outlay necessary. You don’t incur capital expenses. Sure, you might have to pay a bit in the way of startup costs, but you don’t have to buy hardware and software. You simply rent space on someone else’s servers and storage.
Business decision makers love this shift. They don’t need to worry about huge capital outlays, and they know that they’re paying for what they use. They’re not paying for extra hardware that may never end up actually being leveraged to help solve business needs.
When you build your own data center, you have to scale it yourself. Sometimes, you can scale in increments that make financial sense, while other times you have to add more than you might like due to predefined requirements from your vendors.
When you use the public cloud, you don’t have to worry about inherent scaling limits or increments. Remember, you pay for what you use. As your usage grows, so does your bill, but you don’t generally need to manually add new resources to your account. It can happen automatically.
Scalability granularity often isn’t a problem with the public cloud. You grow as you need to. There is no practical limit to how far you can grow as long as the cloud provider still has resources.
Geographic Diversity and Disaster Recovery
Building multiple data centers can be an expensive undertaking, but it’s one that is being executed more and more as companies seek ways to protect their data and ensure continuity of their business in the event of a disaster striking the primary data center. The separate data centers are generally geographically diverse so that a single natural disaster can’t take out both sites at the same time.
Public cloud providers often already have systems that can quickly enable geographic diversity for applications that are already running on their systems. Enabling geographic diversity is often as simple as clicking a mouse button and, most likely, paying some additional money to the cloud provider.
The Public Cloud
It’s hard to avoid the term cloud today. It’s everywhere. For many, the term itself has become synonymous with “Internet” or is just another way to described what used to be called “hosted services.” However, there are a number of traits that make a public cloud a public cloud.
First, in general, public cloud systems are comprised of multi-tenant environments operated by a service provider with the hardware and software located in the provider’s data center. In these environments, the customer may not always even be aware in which provider data center the services reside, nor does the customer have to be aware. The beauty of these systems is that workloads can move around as necessary to maintain service level agreements.
Cloud service providers generally build their systems with the assumption that hardware will likely fail, which means the you, as the customer, can avoid the need to buy expensive failover and availability systems on your own.
For scale, the cloud provider can provide grid-like scalability to great levels so that you don’t need to worry about how to grow when the time comes.
For public cloud, there are a number of pros and cons to consider. On the plus side, cloud will:
- Enable immediate implementation.
- Carry low to no initial deployment costs.
- Provide a consumption-based utility cost model.
- Provide more cost effective scale than would be feasible in a private data center.
However, there are definitely some downsides to cloud as well, which include:
- Potentially unpredictable ongoing usage charges
- Concerns around data location; many do not want data stored in US-based data centers due to concerns around the NSA and PATRIOT Act
- Charges across every aspect of the environment, from data storage to data transfer and more
- No control over underlying infrastructure
- Care needs to be taken to avoid lock-in
The Faces of the Public Cloud
Here is a brief look at the different kinds of public cloud services that are available on the market.
From a customer perspective, software-as-a-service (SaaS) is the simplest kind of cloud service to consume as it is basically an application all wrapped up and ready to go. Common SaaS applications include Salesforce and Office 365.
With SaaS applications, the provider controls everything and provides to the customer an application layer interface that only controls very specific configuration items. Because all of the infrastructure and the fact that most of the software is hidden from the you as the customer, you don’t need to worry about any underlying services except those which may extend the service, such as integrating Office 365 with your on premises Active Directory environment.
Sometimes, you don’t need or want a complete application. In many cases, you just need a place to install your own applications but you don’t want to have to worry at all about the underlying infrastructure or virtualization layers. This is where platform-as-a-service (PaaS) comes into play.
PaaS provides you with infrastructure and an application development platform that gives you the ability to automate and deploy applications including your own databases, tools, and services. As a customer, you simply manage the application and data layers.
In other cases, you need a bit more control, but you still may not want to have to directly manage the virtualization, storage, and networking layers. However, you need the ability to deploy your own operating systems inside vendor-provided virtual machines. Plus, you want to have the ability to manage operating systems, security, databases, and applications.
For some, infrastructure-as-a-service (IaaS) makes the most sense since the provider offers the network, storage, compute resources, and virtualization technology while you manage everything else.
Even though public cloud has a number of desirable traits, there are some harsh realities with which CIOs and IT pros need to contend:
- Security – For some, particularly those in highly-regulated or highly-secure environments, the idea of moving to a multi-tenant public cloud is simply not feasible.
- Bandwidth – Many areas of the world remain underserved when it comes to bandwidth, and companies can’t get sufficient bandwidth with sufficiently low latency to make cloud a feasible option.
- Cost – There may come a point at which cloud may become more expensive than simply building your own environment.
These challenges are reasons that many organizations are turning to private cloud environments.
The term private cloud is often, well, clouded in confusion as people try to apply the term to a broad swath of data center architectures. So, let’s try to clear up some of the confusion.
First and foremost, a private cloud environment generally resides in a single tenant environment that is built out in an on-premises data center, but it can sometimes consist of a single tenant environment in a public data center. For now, we’ll focus on the on-premises use case.
Private cloud environments are characterized by heavy virtualization which fully abstracts the applications from underlying hardware components. Virtualization is absolutely key to these kinds of environments. Some companies go so far as to offer internal service level agreements to internal clients in a cloud-like manner. The key phrase there is “internal clients” — that is the customer in a private cloud environment. For such environments, being able to provide service level guarantees may mean that multiple geographically dispersed data centers need to be built in order to replicate this feature of public cloud providers.
Heavy use of virtualization coupled with comprehensive automation tools reveals an additional benefit of private cloud: self-service. Moving to more of a self-service model has two primary benefits:
- Users get their needs serviced faster
- IT is forced to build or deploy automation tools to enable self-service functionality, thereby streamlining the administrative experience
As mentioned before, many companies want to keep their data center assets close at hand and in their full control, but they want to be able to gain some cloud-like attributes, hence the overall interest in private cloud. As is the case with public cloud, there are a number of pros and cons that need to be considered when building a private cloud.
In the pros column, private cloud:
- Provides an opportunity to shift workloads between servers to best manage spikes in utilization in a more automated fashion.
- Enables ability to deploy new workloads on a common infrastructure. Again, this comes courtesy of the virtualization layer.
- Provides full control of the entire environment, from hardware to storage to software in a way that enables operational efficiency. In other words, routine tasks are automated and repeatable.
- Allows customers to customize the environment since they own everything.
- Provides additional levels of security and compliance due to the single tenant nature of the infrastructure. Private cloud-type environments are often the default due to security concerns.
As with everything, not all is a perfect picture. Private clouds do have a number of drawbacks, including:
- Requiring customers to build, buy, and manage hardware. This is often something that many companies want to reduce or eliminate.
- Doesn’t always result in operational efficiency gains.
- Not really providing what is considered a cloud computing economic model. You still have to buy and maintain everything.
- Potentially carrying very high acquisition costs.
In short, private clouds are intended to have some of the architectural characteristics of public clouds while offering internal clients cloud-like economic outcomes when chargeback processes are implemented. Even if the central IT department providing the service doesn’t really use “the cloud,” as internal clients are able to provision and consume resources on demand — at least to a reasonable point — there is the beginning of a private cloud.
Increasingly, people are choosing both cloud options – public and private – to meet their needs. In a hybrid cloud scenario, the company builds its own on-premises private cloud infrastructure to meet local applications needs and also leverages public cloud where reasonable and possible. In this way, the company gets to pick and choose which services run where and can also move between them at will.
The Intersection of Cloud and Hyperconverged Infrastructure
If you’re wondering what all of this talk about cloud has to do with hyperconverged infrastructure, wonder no more. Depending on the hyperconverged infrastructure solution you’re considering, there are varying degrees of association between the hyperconverged infrastructure product and both public and private clouds.
Everything you’d read so far leads to money. The potential to completely transform the data center funding model is one of the key outcomes when you consider hyperconverged infrastructure. With easier administration comes lower staffing costs. With the use of commodity hardware comes lower acquisition costs. With the ability to scale linearly in bite-size chunks, companies can get the beginnings of a consumption-based data center equipment acquisition model that enables closer to pay-as-you-go growth than traditional data center architectural models. As your environment needs to grow and as users demand new services, you can easily grow by adding new hyperconverged systems.
Agility implies some level of predictability in how workloads will function. Public cloud provides this capability. For those wishing to deploy a private cloud environment, these needs can be met by leveraging hyperconvergence’s inherent ability to scale linearly (meaning, by scaling all resources including compute, storage, and networking simultaneously). In this way, you avoid potential resource constraint issues that can come from trying to manually adjust individual resources and you begin to achieve some of the economic benefits that have made public cloud a desirable option.
Scaling the data center should not result in scaling the complexity. In order to attain the full breadth of economic benefits that go with cloud, you have to make sure that the environment is very easy to manage or, at the very least, that management is efficient. This means that you need to automate what can be automated and try to reduce the number of of consoles that it takes to get things done.
With hyperconverged infrastructure, management efficiency – even at scale – is a core feature of the solution. You are able to manage all of the elements included in the product from a single console and you are also able to apply a breadth of consolidated policies to virtual machines.
Geographic Diversity and Disaster Recovery
Also on the economics front, the value of disaster recovery cannot be overstated. One of the benefits of the cloud is the geographic diversity that can be achieved to protect against natural disasters. With a hyperconverged infrastructure solution that has data replication as a part of the core offering, multisite redundancy capability is baked in as part of the solution.
For those that have opted to build hybrid clouds, some hyperconverged infrastructure solutions can leverage that public cloud deployment as a replication target. In other words, rather than going to the expense of building out a second physical site, the public cloud can be used to achieve data protection goals.
Hyperconvergence and the Private Cloud
Building a traditional private cloud is hard. It takes a lot of work to get all the pieces aligned. Hyperconverged infrastructure can allow you to deploy private clouds in a fraction of the time it would normally take. Everything is built into the individual appliances, including centralized management, data efficiency, replication, and the ability to scale in incremental units. These are core needs in building an agile private cloud environment.