Hyper-convergence has a side-effect on IT operations that needs to be considered before taking the hyper-converged plunge. Hyper-convergence is taking what were separate components of the IT infrastructure and merging them into a single solution. Think of a smartphone. A smartphone takes a PDA, MP3 player and camera and merge them into a mobile phone. Converged infrastructures do the same with IT infrastructure which can create challenges within a highly silo’d organizations.
The CPU, Storage and in some cases network are collapsing into tightly integrated platforms. Examples include Nutanix on the hardware side and VMware VSAN or EMC ScaleIO on the software side. These converged solutions allow for simplified infrastructures that scale by just adding additional blocks of integrated storage and compute.
These solutions compete with traditional compute and storage architectures in both capacity and performance. VMware’s VSAN solution scales up to 4.4PB of storage and two million IOPS. This puts it in the same class of many mid-sized enterprise storage solutions. This is in addition to the compute capacity of up to 32 ESXi nodes.
These are great technical solutions with great technical specs. However the model creates a challenge for operations. Smaller enterprises shouldn’t see much of a challenge. Engineers in these environments tend to work across silos. It’s when these solutions are sold into large enterprises with distinct operations silos. A good place to look at the challenges in a large enterprise is the budget process.
In a normal transaction, a storage vendor can go to the leader of the enterprise storage group and sell a storage array. A server or virtualization software provider can go to the server leader to sell their product. How would you like to be the sales person trying to sell a hyper-converged solution to a large enterprise? Worst still, how would you like to be a server admin trying to buy a solution that includes substantial storage? Who pays for maintenance or how is it divided between desperate groups?
It’s this simple challenge that foreshadows the operational challenges of who manages the design and implementation of these solutions within the operations groups. This is an example of when a great technology is not enough. Operations groups need help defining structures that make choosing and implementing these solutions palatable for large enterprises. Vendors are challenged with helping to make these solutions digestible to static org-charts.
This isn’t the case in every large environment and use case. Some large organizations are able stand up specific groups charged with managing these environments as part of a new project such as a VDI deployment. The challenge comes in trying to retrofit these platforms into existing architectures and their support structure. Before bringing these vendors in-house, you need to have conversations with the stakeholders from departments that the converged platform encroaches. Of course if you can get away with sneaking the solution in and not discussing it with anyone…
I could easily see a customer choosing to go with traditional network, storage and compute to avoid the political complexities of convergence.