Introductory Guide to the Hype and Hope of Hyperconvergence
Check any IT blog in the past few years, and you’re bound to see mention of one of the newest forms of data center technology, hyperconvergence. Choosing a hyperconverged system may be all the rage now, but is it the right solution for your company? To determine if it’s the best course, you’ll need to consider your specific user needs and enlist the help of an agnostic IT Sherpa to guide your implementation choice. This article will provide a primer on hyperconvergence that you can use to inform your decision.
What Is Hyperconvergence?
Hyperconvergence, also known as hyperconverged infrastructure or HCI, or the software defined data center (SDDC), is a data storage concept that combines storage, compute, virtualization, and networking in one unit or SKU to increase scalability, flexibility, and efficiency. Used in organizations ranging from SMBs to Fortune 500 companies, hyperconvergence creates a private cloud, where organizations can store sensitive and mission critical data and environments. Hyperconvergence can provide the storage power of a data center without the costs of space, power, and maintenance. This solution provides you with an easy-to-use data center in your office.
Why Is Hyperconvergence Hot?
Barry Craven, Independent Consultant at XellerateIT
Gartner anticipates the HCI market will form $5 billion or 24 percent of the market by 2019. While IDC says the market experienced 104.3 percent year on year growth in 2016. Some commentators believe hyperconvergence will become the common platform for infrastructure. Ease of use makes it perfect for small and mid-sized businesses. “The good thing about hyperconvergence is it will bring costs down. It will reduce your operating expenses by having less people run it and bring down your capital expenses by having less hardware,” says Barry Craven, Independent Consultant at XellerateIT.
What’s in the Hyperconverged Box?
“By calling it hyperconvergence, it’s a concept, but it’s not a distinct standard, where it has to be this or that,” explains Craven. Nevertheless, HCI has common features. In a traditional data center, the servers use hypervisor to hold the virtual machines. Traffic goes to servers, which then use the storage controller to access the storage area network (SAN) before going through the work of retrieving the data. Hyperconvergence removes the SAN, and the server, storage, and hypervisor that sit above the control layer are one unit called a node. To scale up, just add more nodes. The distributed data plane is managed through one view into all components. A hyperconverged environment uses virtual machines or container functionality. Virtualization requires less equipment in a data center, so hyperconvergence creates the need for even fewer servers—less equipment means more manageability.
HCI features usually include:
- Multisite management for remote boxes
- WAN optimization
HCI packages may also offer:
- Flash data storage
- Guaranteed input/output performance
- Open-sourced products such as a Kernel-based Virtual Machine (KVM) to enable virtualization
- Some hyperconvergence packages have special security features
- Instantaneous backups
With the flexibility of the cloud and the security of knowing your data is on premise, under your maintenance and control, hyperconvergence suggests some obvious uses. A good candidate would be development and test environments, for example, since they may not be suited to the public cloud. Some HCI implementations may work well with large collections of data such as security video or engineering design rendering.
A traditional data storage environment manages networking and storage separately and require balancing, which means separate IT roles, such as storage engineers and network engineers, oversee each aspect. “With hyperconvergence, you can closer to having an IT generalist run it,” Craven suggests. While that means that fewer people are needed to manage data, Craven hasn’t seen mass layoffs. Traditional data centers will persist, especially where companies are vested in huge amounts of data. In these scenarios, the organizations may convert gradually by working on one small use case at a time. “Engineers will go in another direction, such as security, where there’s a growing need,” says Craven.
What Is a Converged Infrastructure?
Hyperconvergence grew out of converged infrastructure (CI), also called unified computing. As with hyperconvergence, convergence seeks to overcome the siloing of IT resources. By creating pools of virtualized servers, storage, and networks, under one management layer, convergence enables sharing of resources across business lines and technologies. Again, similar to hyperconvergence, convergence can make infrastructure management easier. Because convergence relies on physical infrastructure, converged systems are usually purchased for new setups. The individual components of converged systems (networking, storage, and compute) are physical so they can be snapped out for separate purposes.
Convergence vs. Hyperconvergence
A simple way to understand the difference between convergence and hyperconvergence is to remember convergence is about hardware and hyperconvergence is about software. Hyperconvergence adds more abstraction, but also the potential for more automation. Hyperconverged, all elements are virtual and cannot be separated from the package and used for their original purpose because everything is integrated with hypervisor. Scaling up and out is easy—just add hard drives or more servers.
Besides use case, cost is a consideration for choosing between converged and hyperconverged setups. Converged systems often include packages from different vendors. Adding in licensing and support to get the different components to work together, costs can rise. Custom or prefab packages also differ in cost. With hyperconvergence, some costs may be associated with licensing for the hypervisor layer. Despite the integrated nature of hyperconvergence, it is possible to repurpose existing infrastructure for hyperconvergence by using software defined components to efficiently tie functionality together. Hyperconversion may also yield savings by eliminating extra staffing and maintenance.
Some IT experts also recognize another flavor—data center convergence—which is hyperconverged elements added to a traditional data center.
What Is Hyperconvergence Storage?
Hyperconverged storage is software defined storage supplied in a virtual server. In hyperconvergence, the physical box is considered the node in a storage cluster. The benefit comes from being able to pool storage components from each server. Since storage is packaged with other functionality, adding more storage will also get you more compute.
What Is the Definition of Hyperscale?
Hyperscale computing is the ability to adequately and efficiently scale to an increasing number of servers in a distributed computing environment, usually for the cloud and big data.
Hyperconvergence and Cloud Data
The main differences between hyperconverged setups and the cloud are that hyperconverged systems are onsite, and the cloud is offsite. Some vendors offer hybrid platforms that allow you to store mission critical data onsite enabling you to take personal responsibility for security and maintenance.
Cloud packages take away some staffing and infrastructure issues, and offer flexibility and convenience because the day-to-day maintenance belongs to someone else, but it’s not necessarily cheaper. With cloud storage, security is still your responsibility. And, cloud storage has additional privacy considerations. As Craven says, “If the government calls on your cloud host, you might not want them to just hand over your data.”
What Is a Composable Infrastructure?
Composable infrastructures are built off blade technology that allows you to combine assorted off-the-shelf networking, compute, and storage products. Each functionality becomes a service, allowing any application to sit on hardware without extra configuration.
Benefits and Pitfalls of Hyperconvergence Infrastructure
Although hyperconvergence is the it thing in IT, it is not a perfect out of the box solution. Like everything, it has benefits and disadvantages.
- Hyperconvergence offers the flexibility and convenience of the public cloud infrastructure with the control that comes from being on premise.
- One box and integrated technology, means you manage and have visibility across boxes, sites, and systems through one common dashboard.
- Hyperconverged systems include all components in a seamless package, so you can add modules to expand capacity. Scaling is more expensive and harder to do with precision with traditional storage options.
- Add more hard drives to increase storage, or scale up.
- Add more servers to increase compute, or scale out.
- Easier to scale than traditional systems where you needed to gaze into a crystal ball to estimate needs three to five years down the road. As consultant Craven advises, “You can easily scale incrementally as you grow.”
- Better performance because storage is in the same package as the compute power.
- Fast set up of VMs.
- Provides optimization for tier one applications.
- Hyperconverged systems can save power, cooling, space, IT staffing, and licensing costs for recovery and other tools. Some systems bundle a hypervisor, so you don’t have to pay licensing for a separate instance.
- Resilience: The nodes are all connected, so ideally if one fails, the rest take up the load.
- Flexibility: You have a view into all your workloads so you can easily shift them as necessary.
- Scalability: Nodes snap in and out to scale up or down to meet resource needs.
- Cost efficient: SMBs to Fortune 500, can buy as much or as little as they need, when they need it.
- Data efficient: Everything in one box means less need for staff to maintain it.
- Data Protection: Easy to back up.
- Because it’s virtualized, hyperconvergence is not an option if your data application runs on bare metal.
- Some question its actual resilience, and ask how many nodes can fail before major problems occur?
- Although compatibility issues between components from different vendors disappears, you are committed to one vendor for the duration of your product.
- While some express concerns over the security of the unified architecture and its dynamic nature, others point to modern enterprise security methods to match the flexibility of both the system and the incoming threats.
Hyperconvergence Best Practices
The biggest takeaway when considering the move to a hyperconverged infrastructure is to understand your use cases before you buy. A lot of systems are good, but not all systems are ideal for your situation. “My opinion, if you’re going down the hyperconvergence route, is to work with someone who’s agnostic, to make sure you get the right product for what you need,” advises Craven.
In addition, you need to understand your required capacity so what you’re quoted is not more or less than what you need. Craven adds, “You could get something that’s popular and that everyone loves, but it may not be best suited for your needs and what you need to accomplish.”
Explore How Technologies Converge at SHAPE
SHAPE is an immersive event exploring the convergence of technology and entertainment. Discover how the future of content creation and distribution will usher in new audience experiences. Be inspired by visionary speakers. Experience interactive exhibits from promising start-ups and industry leaders at Warner Bros. Studios backlots.
SHAPE is happening June 2 and 3, 2018 in Los Angeles, California, at Warner Bros. Studios.