The level of scalability that can be achieved from hyper-converged platforms is limited by the hypervisor. Every...
hypervisor limits scalability in four areas:
- Number of nodes within a cluster;
- Amount of memory the hypervisor can recognize on each node;
- Number of virtual machines per node; and
- Number of virtual machines per cluster.
These and other limitations remain in effect even in hyper-converged platforms. VMware's limitations don't go away just because the VMware software is running on a hyper-converged system. The same can also be said for Microsoft's Hyper-V, and for just about any other hypervisor.
In some cases, hyper-converged platforms may impose limits of their own. For instance, the servers within a hyper-converged system might be able to accommodate less memory than what the hypervisor is able to support. Similarly, the hardware manufacturer might limit the maximum cluster size in the name of cost or performance, even though the hypervisor could handle a greater number of nodes.
If you need to achieve a greater level of scalability than what hyper-converged platforms support, your best option is to build multiple clusters. Not only does this approach improve scalability, but it eliminates the potential for a single point of failure. Cluster nodes generally share a storage volume and this volume has the potential to become a point of failure for the cluster as a whole unless sufficient redundancy is put into place. Even if failure were not a concern, a storage volume is only able to deliver a finite number of IOPS and excessively large clusters run the risk of overwhelming the shared storage.
Hypervisor vendors fully expect their larger customers to create multiple clusters. VMware and Microsoft have each created tools that allow customers to manage multiple hypervisor clusters through a single interface.
Evolution and implementation tips for hyper-converged platforms
How a hyper-converged platform can replace your SAN or NAS
Reduce strain on a data center with hyper-converged infrastructure platforms
Dig Deeper on Hyper-Converged Vendors and Products
Related Q&A from Brien Posey
Quad-level NAND has its benefits, but is it right for enterprise-level companies? While it can be plagued by performance and durability issues, the ... Continue Reading
Disaster recovery requirements are constantly changing, and a DR plan must do the same. Frequent testing and clear communication can go a long way ... Continue Reading
While cloud DR is gaining popularity, it may not be the right choice for every organization. On-premises DR often comes with more predictable pricing... Continue Reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.