Virtualization has made managing operating systems and applications far simpler, but added its own complexity with...
hypervisors and storage networks. And as the number of virtual machines (VMs) exploded, so did the complexity of the infrastructure required to support VMs.
To tame this complexity and simplify VM management, users have two options: converged vs. hyper-converged storage. While these technologies are meant to solve the same problem, they have very different approaches and suit different environments.
Converged storage system benefits
The first level of convergence is to bundle a complete set of components under one product code and deliver a built platform. The platform is built from existing parts that can be ordered separately. The magic is that the vendor validates the whole platform, and ships it to a customer as a single unit. This also creates one platform to be tested by the vendor for upgrades.
Converged storage system offerings are usually sized to run hundreds or thousands of VMs. Each system is self-contained and managed as a separate unit, and customers can deploy multiple products.
For users contemplating converged vs. hyper-converged storage, there is another benefit to add to the converged storage column -- companies that offer these systems are large enough to sue. For example, Hewlett Packard Enterprise has a market capitalization of $24 billion, and EMC has $51 billion -- before the corporate changes that are underway for both companies. These companies not only build converged infrastructures from their existing products, but they have valuations a hundred times greater than the capital that any of the hyper-converged startups have raised. For a large and risk-averse organization choosing converged vs. hyper-converged storage, there is security in buying from a company that is more valuable than your own.
Hyper-converged storage considerations
The second level of convergence is hyper-converged storage. The big difference here is that the platform is built specifically for running VMs. Hyper-convergence breaks down the silos of storage and compute, converging them into a single building block comprising an x86 server, with local disk and solid-state drives, and some software-defined storage.
A hyper-converged storage infrastructure combines multiple building blocks into a storage and compute cluster. That cluster can be expanded by adding more building blocks, usually without any downtime for existing VMs. These products manage everything required to run the VMs. Hyper-converged storage throws away as much existing complexity as possible and hides the remaining complexity under VM-centric policies.
When deciding on converged vs. hyper-converged storage, a hyper-converged system addresses one of the risks a converged system presents: the risk of business change. A converged system is a large, up-front investment that accommodates the maximum workload envisioned for its three- to five-year lifetime. Very few companies will be able to accurately predict their workload for such a long time -- projects get canceled, mergers and divestments happen, and whole new lines of business spring up. A converged storage system can become a large liability as a business fluctuates.
On the other hand, a hyper-converged storage system only requires an organization to have enough capacity for its current workload. As the workload grows, small building blocks can be purchased to grow capacity. If demand doesn't increase, hyper-converged capacity spending can be deferred. This allows an organization to contain its storage budget and align costs to business value. Even a large and conservative company can benefit from this approach when entering new markets or implementing higher-risk projects.
Organizations deliberating converged vs. hyper-converged storage should consider the low cost of entry associated with hyper-convergence and its simple path to scale as a business grows. And while hyper-converged storage delivers a fixed outcome for a fixed spend just like converged storage, the amount of that spend can be much smaller using hyper-convergence.
Hyper-converged products start to offer more points of differentiation
An essential guide to how hyper-convergence evolved from converged architectures
Converged systems offer validated configurations based on tested components
Which storage method is best? Converged vs. hyper-converged.