Michael Pascoe, managing director of Melbourne-based SI Olikka, hasn’t seen a lot of impact from reference architectures.
“Not much, to be honest,” he says. “They often provide a proof point that the technology works, but rarely do the vendor architectures match the right architecture for a particular customer.” Pascoe is more impressed with hyper-converged offerings from the likes of Nutanix, SimpliVity and Scale Computing.
“In hyper-converged, it probably helps more – because they tend to be newer and less well-known architectures – so the vendors give everybody more knowledge transfer and confidence in the solution. They also have fewer external requirements to make up the solution, so they put more emphasis on the vendor being responsible
for the total solution.”
But doesn’t that remove the SI from the equation altogether? Perhaps SIs should focus on the sales and marketing engine, and rely on vendors and distributors for detailed technical support?
“So be a low value-add sales and marketing company for the vendors? Absolutely not.” Pascoe is emphatic that the SIs play a vital part of the process. “How can you be sure that the vendor solutions really work and are truly good for the customer?” That is the true value of the SI.
Nick Verykios, managing director of distributor Distribution Central, believes vendors have been over-hyping benefits of convergence, just as they did with cloud.
“Resellers, our customers, end up in a state of confusion. It’s an irresponsible approach to the market. It’s like how some vendors say ‘Move everything to the cloud, or nothing.’ ”
Taking a one-size-fits-all approach to solving problems isn’t the way, he insists. “The most responsible solution is always going to be different for different customers.
“The role of the integrator is to figure out how to solve the customer’s problem with available technology. Don’t lose faith in the process,” Verykios says.
Reference case
Sam Pickles, CTO of Aura Information Security, a trans-Tasman security SI, uses reference architectures from vendor F5 Networks to jump-start the process of solving customer problems.
“It’s a useful starting point for experienced hands to guide the design,” he says. “It helps to provide a customer with a level of assurance that the design will be supported by the vendor.
“PODs and appliances are valuable in cases where you have a major enterprise application to roll out at scale. If it’s a shared service across the whole enterprise, then there are different, possibly conflicting, requirements from different areas of the business. You need to make decisions about how to best meet those requirements.”
This is where a systems integrator adds the most value: helping a customer to make good trade-offs without having to be deep experts in the technology options.
Instead of simply plugging in someone else’s hardware, SIs will instead be focusing more on helping to solve business problems, ensuring customers acquire – and correctly configure – the most appropriate technology mix.
“The SIs of the world are changing, and the consultative approach is what they should be doing,” says McCloskey.
Advising customers properly means understanding the unique demands of their businesses. Vendors and distributors are simply too far up the supply chain to have such detailed information.
Systems integrators thus help customers choose between competing solutions to their problems, much as they have always done. But now, the choice is a collection of hardware and software designed to provide business value, like virtual desktops or a software development environment.
For some problems, hyper-converged appliances will be the right choice, while for others, a customised reference architecture will work best.
The role of the SI is generally the same, but has moved up the stack into collections of technology, rather than just isolated pieces.
“A systems integrator is a true advisor, striving for the best outcomes for the customer,” says Pascoe, “They are a value layer between the customer’s environment and the vendor offerings.”
Jargon Buster
Reference architecture A template for how to achieve a certain aim, but with considerable flexibility about the specific components to be used. For example, a reference architecture for a 3-tier web application might specify the number of application nodes, database nodes, or load-balancers, but would not specify the specific vendor of the database or load-balancers.
Converged Infrastructure Multiple components of compute, network and storage are combined to meet a specific computing aim. The components are ‘converged’ together, usually in a single rack (or collection of racks). This is distinct from the deployment of general-purpose infrastructure where servers, networking, and storage are commissioned independently of one another. Examples include the VCE’s Vblock and FlexPods from NetApp and Cisco.
Hyperconverged Infrastructure CPU, RAM, storage and a hypervisor are all ‘converged’ into a single appliance (usually commodity x86-based hardware). To achieve greater capacity or performance, additional appliance nodes are added in a ‘scale-out’ configuration. Examples include Nutanix, SimpliVity and Scale Computing.
Scale-out To achieve more performance or capacity, additional nodes (usually identical, or near-identical to those originally deployed) are added to the set.
Scale-up Additional infrastructure components are added to the system to achieve more capacity or performance, eg, additional disk shelves, or by replacing CPUs with bigger ones. Most traditional systems are scale-up.
-1.jpg&w=100&c=1&s=0)
.png&w=100&c=1&s=0)
