The phrase ‘having one throat to choke’ is a grim way to describe the ideal of having a single supplier who cops the blame when things go pear-shaped. But there are big upsides of converged infrastructure where server, storage, networks, virtualisation and other resources are bonded into one, manageable and holistic solution.
Growing out of this trend is hyper-converged infrastructure – plug-and-play appliances based on commodity x86 servers that pack compute, storage, and/or networking into a scalable module. It is the modular, scalable nature of these blocks that distinguishes this new area from traditional converged infrastructure. (See box, right, for examples of new converged and hyper-converged solutions on the market.)
Modern converged infrastructure appeared in mid-2009 with VCE, the coalition of EMC, VMware and Cisco. The industry has since grown as customers and solution providers gain benefits from vendors’ premasticated solutions and reference architectures.
Accelerating the growth of converged infrastructure are improved performance, perceived lower operating costs, greater IT optimisation, increased automation and simplified sourcing, notes IT analyst firm Gartner.
In June, Gartner estimated the market for “integrated systems” – which is its term for converged infrastructure – would grow more than 50 percent in 2014 over 2013 to reach $6 billion.
Agility and speed to execution are deciding factors for many customers turning on to converged infrastructure, says Dimension Data general manager of data centre solutions Peter Prowse.
“There’s an element of this ‘one throat to choke’,” Prowse says. “The broader motive customers are looking for is this ability to move away from towers of capability. [Traditionally] when you provision new infrastructure for your application, you go to the application, server, storage and networking guys, and that process takes on average 90 days.”
Dimension Data was able to reduce the deployment for ING’s Bank-in-a-Box project down to 10-15 days, says Prowse.
“It was a focus on collapsing their development cycles so they could get a new product into the market. And a big part of that was to provision instances of their environment on the converged infrastructure platform they rolled out inside 15 minutes. That’s the real driver for customers.”
And although integrators may not be spending as much time running benchmarks on hand-cut systems or crawling arcane hardware vendor schematics, that doesn’t mean converged infrastructure has made them redundant. “We do a lot of research and pick leaders in each of the market areas,” Prowse says. “We’re working very closely with Cisco and VMware to deploy software-defined architectures to our customers.”
The shift in technologies that customers prefer is reverberating back through the channel.
Allan King, managing director of Canberra integrator Infront, says resellers need to shift their focus, too.
“We have a very large systems integration practice, but we’ve seen over the past two years our biggest growth in advisory services,” King says. “It’s very much about cloud advisory and workload management and assessment.”
King says that as IT becomes more complex, customers are drawn to the promise of orchestration and automation that converged systems hold out.
“The problem that converged infrastructure is trying to solve [is] getting a single entry point into a block of stuff to orchestrate it consistently. But the maturity of the industry isn’t there to do that and that’s why we look at the Cisco plane to create and build that workflow.”
Begging for blowback
Despite best intentions for a technically superior solution that anticipates the needs of the business, customers and their trusted advisers are dealing with human beings. Collapsing compute, storage and network technologies impacts on the IT staff trained and specialising in those areas. A reseller who walks in to a customer’s IT shop and expects everyone on the floor to buy into the converged systems Kool-Aid from day one is begging for blowback.
“It’s the organisational and cultural change internally that will determine success or failure long before the products you buy,” King says. “That cross-skilling across a data centre practice is absolutely imperative but it’s the thing that happens slowest of all.”
King runs workshops for a customer’s IT staff to explain the changes and inspire them about the outcome. This alleviates specialists’ natural fear that they are about to lose their jobs.
“If everyone understands what we’re trying to achieve and sees value for themselves, what they get excited about as engineers is they will have many skills. So you’re broadening their ability to participate in a discussion by broadening their skillsets.”
The business outcome will often determine whether a converged or hyper-converged solution is chosen. King says that islands of IT can be trouble down the road: “If you choose a product that isn’t integrated [into your roadmap] then your return on investment is eroded dramatically”.
But in some cases, speed to deployment is the overriding factor determining the solution as was the case with Cloud Solutions Group’s implementation of a Nutanix storage-compute appliance for travel broker Webjet.
Webjet had made an acquisition in Singapore, so needed to set up a new co-location that was fast and powerful but had a limited footprint. This drove them towards Nutanix, says Cloud Solutions Group managing director Josh Rubens.
Nutanix dispatches systems with software such as hypervisors and settings already applied so technicians can rack up appliances in minutes. “Nutanix had the box there in nine days,” Rubens adds.
Converged infrastructure accounts for a quarter of the Melbourne consultancy’s business – and this is growing, Rubens says. And he’s “not a massive fan of validated architectures” either, one of the key reasons often given for converged infrastructure.
“None of these things are tailored to customers’ requirements so what environment will you drop a pre-validated box into and not have to tweak?” says Rubens.
Nutanix Australia managing director Wayne Neich says customers are often already well down the path of deciding what they want before they engage a provider.
“If we’re not certain that we can do a good job for a customer, we often put them in training and give them the ability to educate themselves,” Neich says. “They can see first-hand how easy it is to set up, implement and deploy, and how well it works.”
Neich says that as the list of hyper-converged solutions continues to multiply, the sub-sector is “getting more validation every day”.
For instance, Nimboxx broke from stealth mode as the latest startup to unveil a hyper-converged infrastructure solution, presaging what is likely to be a stream of new entrants in the months ahead.
NEXT: New technology coming
Indeed, the second half of the year will see entries from EMC, Dell and Yottabyte and those that were hiding from companies such as VMware and Cisco.
EMC in May confirmed it is building a hyper-converged technology codenamed ‘Project Mystic’, pitting the storage giant against startups such as Nutanix and SimpliVity.
David Goulden, chief executive of EMC Information Infrastructure, told IT solution providers at the EMC Global Partner Summit that it will have a hyper-converged appliance by the end of the year. “It will be very competitive to products starting with N or S,” he said.
EMC majority-owned virtualisation behemoth VMware is also getting into the hyper-converged fray with ‘Project Marvin’.
Cisco appears ready to launch an entry-level UCS server line, giving it a new tool for expanding its server business and possibly entering the hyper-converged market.
Even Dell is getting in on the act with its plan to resell Nutanix’s hyper-converged appliance before offering its own branded solution that ties the Nutanix software stack to a Dell server under an OEM deal.
Dell is also working on ‘Blue Thunder’ to tie technology from partners including Red Hat’s OpenStack and Ceph solutions, Nexenta’s ZFS storage software, and Cloudera’s Hadoop big data software, into converged infrastructure offerings.
For the benefits of automation and orchestration that derive from single-source origins, converged infrastructure of either stripe isn’t for every customer.
“Some customers are very conservative,” Cloud Solutions Group’s Rubens says. “Sometimes you can turn them around, but others only buy IBM or NEC and they will be too hard. [But] if their cultures are innovative like Webjet, those customers thrive on disruptive technology.”
Infront’s King agrees. “We find smaller organisations are more
agile because they usually have asset replacement cycles that are aligned. The bigger organisations have asset cycles anywhere from a one to four-years gap. It takes a brave customer to step into the fold and say, ‘I’ll get rid of it all’; not many customers are in that position.”
Data#3 general manager Laurence Baynham says the channel and customers must grapple with the question of where to break into the upgrade cycle.
“That’s at the nub of what’s going through customers’ minds right now: what is the catalyst?” Baynham says. “It may be a major network upgrade that requires significant spending or it could be storage or compute that serves as a major event. [Customers ask] ‘Do I consider running it independently or do I run it converged?’ ”
Baynham sees a tight link between virtualisation, the cloud and converged systems but cautions that “it’s not a question of buying converged infrastructure and plugging it in… there’s still complexity”.
This comes from balancing workloads and, in a holistic sense, defining where applications and their data should live – on-premise, in a hosted data centre, out in the public cloud or a mix of the above.
It’s in this complexity that arises from a quest for simplicity where a trusted adviser makes their mark, says Corie Marinucci, general manager of HP partner Triforce Services Australia.
“We’ve been able to create a subset of reference architectures around converged systems portfolios, particularly with HP where
we can develop reference architectures for different situations,” Marinucci says.
This includes parachuting converged systems into existing customer data centres, adding to hybrid systems that marry public cloud and on-premise networks and grafting vendor reference architectures into customers’ existing roadmaps, he says.
Sydney-based managed services provider Oriel has also found a way to dovetail hybrid cloud with the allure of converged infrastructure through it’s “your place or mine” promotion. As customers are less obsessed with a badge on the tin, service-level agreements are taking their place.
Oriel managing director Jake Wynne explains that the company is trying to say it doesn’t matter where the infrastructure lives. “Some customers will need to be on-premise for valid reasons. It’s not a one-size-fits-all for this process of converged infrastructure. Everyone is looking for a business outcome, not just holding on to a server and staring at it.”