Going hyperconverged? Don't forget to burst into the cloud

Building a solid team

If the approaches to controlling and integrating these on- and off-premise assets are new, then companies will have to pull together the necessary skillsets to make it all work. That is likely to be politically and logistically challenging, warn experts.

“It has a political effect because you have a different setup from the team’s perspective. The responsibilities change in the IT ops team,” points out IDC’s Nebuloni.

Even the most basic hyperconvergence projects can be political, he warned. Even pooling some data from a customer database still requires you to ensure that a number of people are fine with it, such as the DBA, the security officer, and the application owner. “You need to go through a lot of hoops to ensure that it works.”

The skills required to manage hybrid cloud solutions in a managed environment are also unique, and may well develop separately to those in the rest of the infrastructure. Analysts suggest that most hyperconverged kit today is being bought on a per-project basis, which can create a natural divide between the people managing the hyperconverged kit and those toiling away on conventional general purpose tin.

“You have to be treat this as a discrete operation to start with. It does need to treated outside the mainstream,” says Bob Plumridge, chair at SNIA Europe, who gets feedback from his members about how they’re approaching hyperconvergence. “It’s different enough that it requires people who are at least initially dedicated to this, and who have the time to investigate which applications work best and which don’t work so well.”

Inside those discrete teams, IT departments will need workers with ninja networking skills to manage the different network layers, says Nebuloni, and your team should have a robust set of virtualization skills, too. This doesn’t entail a team filled with specialists though, adds Plumridge. The lure of hyperconvergence is that it hides the complexity from you. Staff with specialist knowledge in networking and virtualization may be necessary to set up more complex hybrid hyperconverged environments and to troubleshoot situations – ideally with the vendor’s help – should things go wrong.

“But instead of everyone, it’s maybe one or two people,” he says. “And you see more people becoming generalists. They can deal with the servers, and the networks and with application performance and layout. They’re more able to look at multiple areas of the IT infrastructure than they did two or three years ago.”

That’s a function of the programmatic environment underpinning most hyperconverged environments, Plumridge adds, and that’s an area where specialist skills will definitely surface. Peter Duffy, CTO at capacity planning services firm Sumerian, says that the skills required to set up and administer these boxes – especially in hybrid environments – are similar to those found in DevOps teams.

“One of the DevOps skills is writing scripts to automate various processes so that I’m not having people around to do things all the time,” he says. “You’re writing scripts to talk to manufacturer’s APIS and bring back that data and push it through another set of APIs to get it into a vendor’s tool.” That management automation is a big part of the hyperconverged world.

As these hyperconvergence projects mature, they may need service management skills, suggests Duffy. He envisages a management layer that translates between business service requests and the underlying hyperconverged infrastructure, so that business service requests can be mapped to the appropriate services in the hyperconverged box.

“It also goes in the reverse,” he says. “If I detect an issue in my infrastructure, how do I then navigate that back up through this wave of relationships and work out which services will be affected by that particular issue?”

Mapping the business services you’re providing to the underlying hyperconverged infrastructure – especially when it’s part of a hybrid cloud environment – is a key part of that skill set, says Duffy. Understanding what another hundred users will do to a hyperconverged box – and what you might have to offload to the cloud - will be an area of high demand as the concept matures.

This IT service management layer requires a dedicated skill set and the ability to tie ITSM and workflow automation into the hyperconverged box’s management software. It’s likely to become more prominent in these teams as hyperconvergence evolves and supports a wider variety of workload. Over time, it will become easier to connect these systems to public clouds, and the software designed to manage both on-premise hyperconverged environments and public cloud services in concert will evolve. As they become more confident in the capabilities of hyperconverged systems, companies will find value in specific cases where it makes sense to have failover or specific back-end services that can scale in the cloud, supporting these tightly-integrated architectures on their premises.

Before this happens widely, we’ll have to see hyperconvergence mature. As a relatively young technology, it’s bound to evolve in the next few years. If it wants to expand its scope from discrete, project-focused workloads into more general-purpose computing environments, it’ll have to – and the teams that support those deployments will need to evolve along with it. ®