Converged and Hyper-Converged Infrastructure. In the last week, VCs have funded six new HCI startups, four major vendors have launched new HCI products, and customers have deployed millions of VMs on it (OK, none of that is true, but given the enormous amount of hype, you considered buying it, right?).
With all the hype about converged infrastructure, we’re focusing on the wrong things:
- Mistake: Convergence = Product or Architecture.
- Reality: Products are over-hyped right now, but convergence as a disruptive trend is being underestimated.
- Mistake: Convergence = combining existing parts of the infrastructure.
- Reality: Convergence = transforming the relationship between business applications and information.
- Mistake: Convergence = better packaging of IT infrastructure.
- Reality: Convergence = mechanism to drive new consumption models.
Last time, we talked about convergence as a trend. This time, we’re going to talk about how most of the approaches to convergence are wrong.
We’re Looking at the Small Picture
Most convergence projects improve short-term efficiency, but do not change the relationship between the business and IT. While simplifying IT processes is valuable, to remain relevant in the “on-demand cloud-centric” world, IT teams need to do more than streamline existing operations. The short-term focus of convergence comes from the natural biases of both the teams creating converged products and those consuming them.
In the short term, many converged products can simplify the management of our products for backup admins, storage admins, server admins, etc. For example, integrating protection software and hardware simplifies backup management and reduces the workload of the backup team. Integrating hypervisor, server and storage simplifies the provisioning of production VMs and reduces the conflict between the VM and storage teams. Incremental steps like these simplify and automate existing functions, so that IT can deliver the standard services to the business with fewer people.
Streamlining components of IT infrastructure does not change a business user’s experience with IT. One customer unwittingly provided a perfect case study. Most converged infrastructure products focus on simplifying the provisioning and management of production VMs. This customer boasted that he could provision a VM in under a minute, when it used to take six hours! Then he admitted that provisioning the production VM is only “90% of the work”. That final 10% just required him to integrate the new VM with his organization’s independent solutions for operational & disaster recovery, archival, test & development, collaboration, and distributed data access. Until that protection and data management is set up, that new VM is not fit for running a production application. In his case, that “final 10%” takes a week. In other words, to the end user, his total time to provision a new VM remained exactly the same.
Other types of “short term” convergence also struggle to dramatically improve the business teams’ experiences.
Converged backup products simplify the configuration of a separate, standalone backup environment. This reduces the effort and complexity of managing backup infrastructure… and nothing else. It doesn’t make data recovery faster, connect better with the production environment, or enable the user to recover their own data. Again, it’s a useful step for reducing IT operational expense, but no more.
Converged infrastructure application stacks simplify the end-to-end configuration and management for that one application… and nothing else. That application lives on its own island, but doesn’t connect with the other services and tools that comprise the business application. In fact, by separating itself, the converged application silo makes it more difficult to provide end-to-end services for the application the business cares about. Nobody knows exactly how to connect that island with the rest of the infrastructure.
Why are we Thinking Small?
Why do so many convergence products think small and focus on the short term? Ego and Fear.
Most product teams suffer from “Center of the Universe” Bias. We believe that our product is the most important thing in a customers’ environment. After all, each of our features is indispensable, while the other parts of the infrastructure (i.e. those we don’t build) are just commodity. Therefore, each of us builds converged products that center around our core competency. Since we can’t all be the center of the universe, ultimately, most of us are wrong.
Most product teams and customers (OK, most people) fear change. Product teams know what their target users expect and want. It’s difficult to build a product for an IT role that doesn’t currently exist. IT organizations hesitate to completely change their organizations. Even with the threat of cloud outsourcing their functions, they are afraid to change. They wait for their incremental optimizations of existing people, process, and technology to save the day. Some wait for the magical product that will save them. In either case, they’re waiting in vain.
Next time, we’ll talk about how Converged Infrastructure (on a big scale) has the potential to connect IT to the business. Short-term converged infrastructure products and projects deliver value. They optimize existing IT functions.
While short term convergence isn’t the big answer, there is hope. If these steps are part of a larger initiative to free up time and energy to tackle Converged Infrastructure on a scale that affects the business, then you aren’t just waiting and hoping. You have a plan.
Converged Infrastructure must strive for more than optimizing localized IT functions.
Stephen Manley @makitadremel