No, we didn’t get Elvis Costello for the Software-Defined Data Center (SDDC) Symposium. Hopefully, the audience found more benefit in a day’s worth of discussion about what comprises the SDDC and, maybe more importantly, why anybody should want one.
To that end, the vendors presenting Tuesday morning were pressed for use cases — examples where something like an SDDC is starting to take shape or is already in place. One common motivation that emerged was segmentation of the network.
Such a thing is useful when dealing with the new found mobility of IP equipment, for one thing, said Don Clark, U.S. director of business development for NEC. In this case, “mobility” refers not to cellphones but to mobility — hospital equipment that’s moving from floor to floor.
“Traditional networking requires the isolation of different departments across a hospital infrastructure, which means manual configurations on a device-by-device basis. What Kanazawa [University Hospital] was able to do was eliminate the traditional firewall architecture and implement a tenant-based isolation infrastructure,” said Don Clark, NEC’s U.S. director of business development.
Embrane and Brocade both brought up isolation as well, but from a more conventional point of view: the need to separate customers’ data and provide flow-specific firewalls — things they’ve had with the physical network but are in danger of losing when workloads get spread out to the cloud. “The cloud service providers don’t give you that much control [by default], because they’re being as generic as possible,” said Kelly Herrell, vice president of software networking at Brocade
Making SDDC Worth It
To really draw customers into the SDDC idea, though, it’s going to take more than individual use cases. What customers want is alpha — a measure of reward that far outweighs the risk, Herrell said. “If the return is just balanced with risk, customers are going to be tepid,” he said.
To some customers, the SDDC does indeed have that return, which is why we’re talking about the concept in the first place.
“It’s not a case of me trying to go in there and sell it to customers that are unwilling. The customers are demanding virtual appliances that can get them the cost benefit and agility,” said Geoff Arnold, a cloud networking architect at Brocade.
The problem is that this expands the scope of the networking team, and not every organization is ready for something like that.
Marco Di Benedetto, CTO of Embrane, endorsed a phased approach when it comes to altering operational procedures and personnel. A massive change is the goal for the long-term, but in trying to sell Layer 4-7 services, he’s finding he has to work alongside appliances and the operational traditions that go with them.
Bringing up the case of Ryan Labs Asset Management, one of Embrane’s announced customers, Di Benedetto noted Ryan wasn’t able to “deliver a single software-defined, optimal experience” in one blow.
“Whatever you start doing to get to the vision of being able to take individual applications and run each application independently to start brokering these applications … this vision can only be delivered if you start with baby steps,” Di Benedetto said.
Along the way, the SDDC could be constructed piecemeal — doing the virtualization of, say, storage first and saving full network virtualization for later, he said.
Most of the other speakers disagreed, contending that an overhaul of processes — ripping the Band-Aid off quickly — is a better approach. In part, that’s because politics and “turf wars” are what’s causing the reluctance to move to SDN, said Inder Gopal, an IBM representative who’s chairing the OpenDaylight Project.
“At some point, you have to say, ‘What should my organization look like?'” Arnold said.