An unexpected result of deploying a hyper-converged infrastructure is that it changes the purchasing cycle for some data center components. Infrequent but large purchases give way to frequent, smaller buys that reduce the amount of resources that don't immediately deliver business value. What was once a large hardware replacement project in a traditional environment becomes a nondisruptive rolling upgrade in a hyper-converged data center.
In addition, many customers don't realize that rolling upgrades can remove the impetus to re-evaluate their current setup. This can result in hyper-converged infrastructure customers staying with one vendor for a long time.
Traditional data center cost model
Traditional infrastructure is purchased in large pieces that are replaced every three to five years. In many cases, the budget allows only one piece of the infrastructure to be substituted each year. For example, if the storage array is changed one year, next year's budget may be assigned to swap out a compute cluster. Replacing a storage array means moving every virtual machine from the old array to the new one. Switching compute nodes is a little less effort, but the project can run for a month or more. Silo-based infrastructure tends to be difficult to grow over time, so excess capacity is bought to cope with anticipated demand. If the demand grows at an unanticipated rate, the infrastructure may not serve the business well.
Cost model of a hyper-converged data center
Enterprises can buy enough hyper-converged infrastructure for near-term needs and, as workloads change, add nodes to an existing cluster, usually without downtime. Because new nodes are easy to buy and add to existing hyper-converged data centers, this pay-as-you-grow model aligns the cost of infrastructure with the value delivered to the business.
There are two significant changes to updating a hyper-converged data center:
- Compute and storage are bought together, rather than as separate purchases.
- Purchasing can be spread over time.
Analyst Ben Woo explains what enterprises should consider before deciding to hyper-converge their data centers.
Hyper-converged infrastructure nodes are also replaced every three to five years, but you are less likely to replace every node in one big upgrade because they are purchased a few at a time. The majority of hyper-converged infrastructure platforms allow virtual machines and data to migrate easily from the old nodes to the new. Businesses that are highly seasonal, for example, could buy additional nodes before the season and retire the oldest ones when the season is over. This is another way the node-based architecture of a hyper-converged infrastructure enables simpler alignment of cost to business value.
While a hyper-converged data center can change the way infrastructure is bought, there is no trigger to re-evaluate your choice of a hyper-converged infrastructure vendor. You should review your organization's hyper-converged infrastructure every few years to ensure it is the right platform for your current needs.
Be aware of vendor-specific hyper-converged scalability limitations
Users approve of hyper-converged infrastructure cost model
Evaluating hyper-converged architecture vs. SAN cost