Converged systems remain one of today’s most talked-about areas of enterprise technology. While there is a lot of buzz around hyper-converged infrastructure from fast-growing vendors such as Nutanix and Simplivity, converged systems from established players – such as VCE’s Vblock or the Cisco and NetApp FlexPod – remain a focus, especially at the big end of town.
Converged systems sit between the tightly integrated, appliance style of hyper-converged systems at one end and do-it-yourself collections of individual components at the other. CRN spoke to vendors, resellers and customers to find out how they’re using converged systems, and why.
Ease of use
The major reason for choosing a converged system over assembling something yourself is ease of use. Figuring out the full list of components to connect, and which ones are compatible with others, is a Herculean task. By purchasing a converged system, you outsource most of the risk of incompatible parts to the vendor, which has presumably tested the configuration fairly well in order to certify it.
If nothing else, you have a single phone number to call if things come a cropper. Many vendors like to lean on this “one throat to throttle” model of support when selling a deal. Chris Trevitt, EMC’s former general manager of channel sales, Australia and New Zealand, cites a dairy customer from Western Australia that chose to run its whole business on VCE’s Vblock stack. “They wanted technology that they trusted from brands that they recognise, and they wanted something that they could really have one person manage.”
Converged systems provide additional flexibility when compared with the far more tightly coupled hyper-converged systems. “You can do a thing called selective scalability,” says Len Findlay, chief executive of Sydney-based integrator VMtech, an award-winning NetApp partner and winner of the CRN Fast50 Editor’s Award last December. “I might need more compute, but I don’t need to grow the storage.
“You can build it for the base environment that you know you have today, then later – say you need to add VDI, which is a completely different workload profile – you can grow the environment to add the CPU and RAM that VDI needs, without adding storage,” Findlay adds.
Cost savings
Converged systems are generally priced higher than comparable hyperconverged systems or the equivalent collection of raw kit, but every vendor and reseller we spoke to was keen to challenge CRN’s assertion that converged systems are expensive.
“Converged systems are perceived to be expensive when compared simply as ‘tin for tin’,” says Darren Adams, Avnet Technology Solutions general manager, Australian & New Zealand. “However once viewed in terms of integration, implementation and management of the systems, cost benefit weighs heavily in favour of converged systems.”
Peter Prowse says: “Dimension Data can deliver a private cloud on a Vblock in 90 days, and that’s up and running production workloads.” It’s this reduced time-to-value that provides much of the savings, because customers generally struggle to manage it themselves.
Further savings come from the reduced complexity of systems administration when compared with the traditional build-it-yourself approach. “The upfront cost is just one aspect of the total cost of ownership,” says Trevitt. “A second aspect is cost of management.”
Dimension Data’s Prowse agrees. “Really, it’s actually about the toolset that manages the infrastructure and how you deploy and maintain that toolset.” He argues that to focus on the infrastructure is to miss the point: “It’s the stack that makes the difference.”
What isn’t clear is if vendors are using a reasonable comparison. Replacing any technology that is five or 10 years old with newer kit is likely to save money in one way or another, so it can be hard to work out if the TCO savings are real. If lifetime cost savings are a big part of your justification for purchasing a converged system, it’s probably worthwhile to get an independent assessment of the way the cost savings are being calculated to make sure.
Next: The trade of in flexibility
Flexibility
Converged systems represent a trade-off in flexibility. When you choose a converged system, you lose the ability to make fine-grained choices about components that you have when building something from individually selected components. But you gain a set of validated components known to work with each other, which are less likely to encounter nasty ‘gotchas’ down the road.
Marketing and advertising company STW Group chose converged infrastructure, specifically FlexPod from integrator VMtech, for precisely this reason. “We wanted something that wasn’t as prescriptive as some of the other in-a-box solutions,” says STW Group CIO Tom Ceglarek. The company installed FlexPod infrastructure a little over five years ago and has grown it to suit their evolving requirements.
It was also important to STW Group that the system was very much integrated when it came to the support arrangements. “There’s people willing to stand behind this combination of products and take responsibility,” Ceglarek says.
Who’s using it?
“If you look at the telco industry in Australia, you’re going to find several telcos using Vblocks as the foundation of their private cloud capability,” says Trevitt. Telstra offers private cloud services on Vblocks out of NextDC.
System integrator Dimension Data also uses Vblock to power its own private cloud for customers, even if the company implemented a FlexPod-based private cloud for ING Australia.
The bank’s choice also touches on the distinction within the converged systems space between an all-in-one stack from VCE and the blueprint provided by Cisco and NetApp. “While FlexPod is absolutely a converged system, it’s still a reference architecture rather than a pre-configured or pre-engineered solution,” Prowse explains. “ING needed some customisation for some DevOps and application requirements they had.”
Who profits?
Are converged systems good for end customers, or do most of the benefits flow to the channel?
“Customers should really only care about the outcome they’re getting,” says Peter Prowse.
“Most of our conversations with customers are around SLA and legal constructs,” says VMtech’s Len Findlay. He says customers only need to care about the gear underneath so that they believe the agreed SLA is achievable.
“We care about the fact that it’s FlexPod because we’re trained to be able to deliver that technology very, very well.”
The move to using pre-integrated hardware stacks like converged and hyper-converged infrastructure means that channel partners are having to become more skilled in total stacks of infrastructure and software, as well as the business problems they’re helping to solve.
It’s not about the tech
“There’s a great opportunity to wrap high-value services around converged infrastructure, to transition their businesses from comparatively lower-value rack, stack and integrate activity, and move it more to managed services or private cloud capability,” Trevitt says.
“Customers are asking partners more and more to focus on the outcomes they can deliver and are less concerned with the procurement function that resellers used to provide,” he adds.
Partners should be making sure they are able to continue to provide services that customers value, which increasingly have nothing to do with racking and stacking gear.
To remain a valued partner, you must start talking to customers about their business outcomes and helping to broker a relationship with the vendors who best help customers to resolve their business problems.
What are converged systems?
Reference architecture A reference architecture defines a set of components – compute, storage, networking and software – that can be combined to build a system. The customer or systems integrator is responsible for procuring and assembling the components. Example: NetApp FlexPod
Engineered system A set of compute, storage, networking and software components is assembled by the vendor and delivered as a unit, ready to be installed in the customer’s data centre. Example: VCE Vblock
Hyper-converged system Compute and storage pre-configured and purchased as ‘nodes’ that are connected over a network. The network components are usually purchased and installed separately and do not form part of the node configuration. Example: Nutanix, Simplivity
A brief history of convergence
The concept was sound, even in the early days. New hardware and software helped bring the industry’s top vendors to the table as new offerings hit new performance benchmarks.
2008 - Exadata
Arguably the first converged system was the Sun Microsystems (now Oracle) Exadata database machine launched in 2008 as the first of its “engineered systems”. The Exalogic product line was introduced in 2010, positioned as a more generic compute platform than the database specific Exadata line.
2009 - Vblock
VCE was formed in November 2009 as a joint partnership between VMware, Cisco and EMC to develop cloud computing platforms originally known as Vblock infrastructure packages. Cisco and EMC announced a joint venture at the same time called Acadia, which would build Vblock infrastructure packages for customers. Acadia was combined into the VCE entity in January 2011. In October 2014, EMC acquired 90 percent of VCE, while Cisco retained 10 percent.
2010 - FlexPod
NetApp launched its FlexPod reference architecture in 2010, in partnership with Cisco. FlexPod was widely seen as a competitive response to EMC/VCE’s Vblock.
2011 - CloudSystem
HP (now HP Enterprise) launched its CloudSystem platform in January 2011, which was merged into the Converged Systems portfolio in April 2013.
_page-0001.jpg&w=100&c=1&s=0)
-1.jpg&w=100&c=1&s=0)
.png&w=100&c=1&s=0)
