If you're confused about the differences between hyperconverged and converged systems, this product comparison might help you decide which approach to take.
Enterprises that leverage SAN-based storage rely on LUNs to set the class of service (CoS) for storage services. However, many of the hyperconverged infrastructure (HCI) vendors have been discussing the operational benefits of HCI over traditional SAN models. The case for HCI is that LUNs are outdated units of management for enterprise storage. A common mantra from Nutanix has been "no more LUNs."
To help get a sense of the differences between the two operating models, I deployed a converged solution --VCE's Vblock -- and a hyperconverged solution -- EMC's VSPEX BLUE. Let's see how they stack up.
Vblock, the converged approach
The concept of the converged system is to bring together the best individual infrastructure components from separate vendors into one engineered solution. The entire platform is supported by a single vendor.
Vblock seems to be the standard by which every converged vendor aims to imitate or compete. Analysts believe converged systems such as vBlock have large profit margins. The highly engineered nature of the platforms fuels and justifies (some argue) high-profit margins.
In the case of Vblock, VCE combines its storage platform with Cisco servers running VMware vSphere and Cisco networking. EMC's wholly owned subsidiary VCE sells, installs, and supports the entire vBlock infrastructure.
A typical use case is the infrastructure for a single Tier 1 application. Organizations looking to keep mission-critical applications up and running on highly available infrastructure may look toward a converged system. Another use case is a reference pod for a data center refresh; it's common to see vBlock as the reference architecture for building and expanding an entire data center infrastructure.
VSPEX BLUE, the hyperconverged approach
VSPEX BLUE is based on the EVO:RAIL program for OEMs, which was introduced at VMworld 2014. OEMs could take a reference architecture that include a 2U 4-node appliance. Similar to Vblock, VSPEX BLUE provides compute, virtualization, and storage in a highly engineered solution. Customers must bring and support the network stack.
Common use cases are VDI and Tier 2 applications as well as remote offices. EVO:RAIL and, therefore, VSPEX BLUE, doesn't scale beyond 8 appliances/32 nodes per cluster. It's not a likely candidate architecture for large data center deployments due to the lack of cluster scaling.
Deployment differences between Vblock and VSPEX BLUE
VCE advertises that a Vblock deployment will take as little as one week; if you consider the system's complexity, this is a short time span. I've overseen projects in which installing Vblock's storage option VMAX took weeks to design and implement. The idea that storage, Cisco UCS servers, Cisco Nexus networking, and vSphere can all be stood up in a week is impressive. Organizations selecting this architecture will likely have an organizational model and workflow to administrator and leverage a Vblock.
VSPEX BLUE only takes hours to rack, stack, and configure. With the quick installation comes a different administrative experience. Designed for smaller environments, EVO:RAIL customers may not have dedicated infrastructure teams for network, storage, and compute. What does this mean for the everyday experience? Engineers lose some flexibility. For example, IT admins may not be able to tune storage for applications such as transactional databases that may benefit from different tiers of storage.
The market is filling in these empty spaces, and I am expecting to see a platform from EMC that sits between VSPEX BLUE and Vblock. Additionally, solutions from Nutanix, Simplivity, and Maxta overlap between EVO:RAIL and converged solutions.
We want to hear from you
What do you think about converged infrastructure? Does the installation effort factor into your purchasing decision? Tell us in the discussion.
- Nutanix ups VMware rivalry with own hypervisor, management tool (ZDNet)
- VCE launches new converged systems, sets path for future (ZDNet)
- Meet hyper-converged infrastructure needs by training engineers to be architects
- Don't invest in a hyper-converged solution before defining your target use case
- Hyperconverged technologies: The new must-have for very lean IT departments
- Hypervisor-converged systems: Pros and cons to consider before taking the leap
Note: TechRepublic and ZDNet are CBS Interactive properties.