26 mai 2023

13 min

BTEX 2023 : How to Navigate the Multicloud Universe

What should a hybrid cloud evolutionary journey look like and what are the challenges and opportunities to consider?

BTEX 2023 Cloud Panel

Mass adoption of public cloud for all things, including data storage, applications and infrastructure, was for a considerable time a computing endgame envisioned by many businesses and organizations.  

Times have changed. Organizational demands, new business models and never-ending technology innovation brought about a realization that one way isn’t the only way – nor should it be. Hybrid multicloud – a combination of on-premises, public and private clouds – recognizes data and processing capability must reside both on and off premises and perhaps even with more than one cloud service provider.  

A panel discussion at CDW’s recent Business Technology Expo in Toronto shared the views of industry professionals around the topic of “Navigating the Multicloud Landscape.” Questions related to what a hybrid cloud evolutionary journey might look like and should consider were posed and answered.

“I think what customers, businesses and organizations have really come to find out is the (on-premises) data centre still has relevancy,” said CDW Canada’s Hybrid Cloud Principal Technology Strategist, KJ Burke. “The edge still has relevancy. The public cloud obviously has relevancy.”

So how might organizations move forward with a hybrid multicloud strategy and what are some of the key issues to consider? Below are panelist responses to queries posed by CDW’s Burke.

Question: How should organizations look to build the cloud operating model in the data centre?

According to Dell Technologies Advisory Solutions Consultant, Briar Main, customers everywhere seek to understand what benefits the cloud building blocks have truly brought to business and how those concepts can be taken and conceivably brought back into on-premises data centres.

“What we’ve learned from using the cloud is that it provides an ability to simply and quickly spin things up – such as containers and automation,” she said. “It has proven to be a benefit. The question now is, how do we bring that back and deliver that ease of use on-prem, but in a secure and controllable way?

“How do we take workloads and connect them between both on-prem and public cloud instances, at the edge, et cetera?” Main added. “You want to make sure that you're building flexibility within your organization.”

She explained that site reliability is vital when looking to move data resources on-premises. When you're looking to create a second copy of your data, you want to understand where that sits and how it is replicated. You also want to consider how to build in data protection elements and determine what types of data should reside where.

“Data is the most crucial element and the most important part of building a business and continuing to move forward,” Main said. “So, you need to make sure you’ve covered all aspects. I think the ability to either replicate to a public cloud or create replication between public clouds is really where we're going to see workloads moving in the future.”

Question: What are some of the drivers that cause organizations to re-evaluate data centre capital expenses and big purchases?

According to HPE Canada Solution Architect Director, Igor Samuk, flexibility is among the most important considerations for data centre investments, since it has completely changed and now extends out to the edges, core and into the cloud.

“We have to think through that,” he said, explaining the customers he works with want data flexibility and the ability to move it around. Flexibility also includes bringing data, systems and applications back on-premises, if necessary. Much of early cloud adoption was CEO driven, but we’ve since learned that “not all workloads apply and sometimes they need to come back. So, decisions have to be made there,” Samuk said.

“The nature of the data centre is also changing dramatically. We have different tiers for availability aspects. That must now include more of an edge-focused model as well.”

Burke made the point that many organizations are placing greater focus on costing per project, which means cloud and data centre investment decision-makers are not necessarily IT operations teams. They can be project teams, developers and business executives, he said, adding that IT investment is not simply “a bucket of funds that IT distributes.”

Samuk agreed, adding that customers are also asking for utility IT investment models, such as managed services, where they pay for what is used or needed. He added that managed services will also help organizations address skills gaps and lack of available IT talent to administer and run data centres.

“I'm seeing our customers say, ‘I'm going to hire for where I'm going, not necessarily for what I have,’” he said.  “Giving customers that flexibility where (a vendor) can say, ‘Hey, we're going to build this model for you and you can bring in services as you need them,’ addresses some of that out-tasking element.”

Question: What challenges do customers face from a networking perspective to successfully shift away from cloud-first and focus more on hybrid multicloud?

Brit Hennig, whose responsibilities for Cisco Canada include networking and cloud architectures, says there are three perspectives to consider: improving application, operational and cost performance. He outlined a scenario of microservices running in containers, and rather than all running on the same server, they may be running on adjacent servers or elsewhere in a data centre.

“Some might be running microservices in public cloud or in a private data centre and that introduces a challenge in terms of ensuring application performance,” he said. A traditional architecture utilizes traditional enterprise networking multiprotocol label switching (MPLS) as well as the internet so that applications might run on public cloud or in an on-premises data centre.

“They might be running on the edge. Your users might be in a campus, in the branch, at home or at the local Starbucks,” Hennig said. “It’s many-to-many connectivity that you worry about. You can introduce SD-WAN (software-defined wide area network) solutions that can connect your branch, your campus and even your end users, then connect to the public cloud, your private data centre and potentially to SaaS applications as well. That allows you to manage everything from a common dashboard with a common security policy and helps with your operational challenges.”

He added that more mature SD-WAN solutions can be connected into middle mile (the connections between service provider networks and local area networks) providers who would aggregate traffic. Mid-sized companies can then connect into the networks of these providers through their SD-WAN solution and take advantage of cost-effective internet. However, for reasons of compliance, security or performance it may be necessary to retain an MPLS network underneath, and in that case, application steering to that MPLS network would be necessary, Hennig explained.

As for the networks within a data centre, Hennig said the challenge here is often dealing with a multivendor environment, which may result in operational silos.

“That's what we're seeing with many customers – (a division of) cloud teams and legacy teams. It doesn't have to be the case. There are solutions…that allow you to implement a security policy…that can be extended on top of the cloud network,” he said. That creates what Hennig described as “a common playing field” across those networks.

“They are learning about one network…and integrating application performance management tools into one network tool set. That allows them to obviously save costs and move applications easier.”

Question: What are some of the approaches that customers need to look at to move forward in that multicloud network?

Hennig likewise addressed this question, suggesting that strategically, organizations need to be wary of building silos. Assume you will implement a cloud operating model across the public cloud and your private infrastructure, he said, adding that doing so will modernize your private infrastructure and ultimately how you do business.

“There have been major advances in cloud architectures, particularly at the hardware level and all the way up through to the management stack,” Hennig said. “Customers can take advantage of that today.”

On a more tactical level, Hennig said there has been a “kind of religion” around moving applications to the cloud. However, before doing so, it’s vital to know your applications and whether moving them makes sense and is the right decision.

“There are great tool sets that allow you to know what resources your applications consume, where your users are and how they're consuming those applications,” he said. “That will allow you to make smart decisions about which applications need to move, which applications need to be modernized and made cloud native, and which applications…to park on an upgraded server.”

Question: Abstraction continues to be important and when you look at abstraction and the innovation that it is driving forward in the hybrid multicloud, what is it and how is your company approaching it?

Stephen Akuffo, Partner Solution Architect for VMware, said his company is targeting storage and networking innovation on the application programming interface (API) service layer, because a common service layer that everyone understands and is able to work through is needed.

Abstractions that will matter are those that take away business processes and make them technology problems,” he said. Deciding whether to place workloads on a private cloud, public cloud or anything in between today is not just a technology decision problem, he added, explaining it should also consider if an organization has the right IT skills, abilities and capacities.

“(A company might say) ‘We have the skill sets to manage this particular workload real well in-house so that's what we're going to do. And for that, the things we don't know how to do well in-house, we'll get some managed services, we'll get some SaaS applications,’” he said. “Those are the abstraction layers that VMware is working towards.”

“How do we accelerate the multicloud transformation? Not just by focusing on what is a virtual machine and what is a container, but (also considering) what is the right place to do the right job,” Akuffo added. “VMware has been – as our CTO likes to say – ‘Really good at making things that don't work well together, work well together.’ And that's the abstraction that we have focused on.”

Question: What is the evolution for organizations as they look to Infrastructure as a Service, Platform as a Service, Software as a Service, and how has that changed over the last year or two?

Neil Thompson, Partner Sales Director for Microsoft Canada, said he believes there are three main pitfalls that many customers risk falling into when considering the adoption of as a Service offerings. Celles-ci comprennent :

Uninformed choices

Thompson says many customers have described to him what they think the cloud is, but their definitions are often outdated and may not apply to their particular industry. You need to have “a real picture of what the cloud is, not a caricature or something that's out of date,” Thompson said. Without a clear understanding, the conversation gets pulled all over the place. “So just round it on facts – objective-provable, demonstrable facts, not just marketing material.”

Outdated investment considerations

Thompson said he believes there is old thinking around investment models such as return on investment (ROI) and net present value (NPV) that don’t apply to many of today’s as a Service value proposition scenarios and “often sound like CFO dialogue from the 70s.” As a Service differs in many ways from traditional in-house technology provisioning models of build, own and operate. There are many considerations and values around as a Service that were not part of traditional ROI models and don’t fit into old definitions and ways of thinking. 

Limited scope

Sometimes when looking to purchase as a Service offering, organizations might consider too small a perspective. Rather than examining a complete picture or set of challenges, they may focus on one element or immediate need and opt for a partial fix. It might be a decision based on a limited budget or belief that addressing one glaring challenge solves everything else. But one particular facet may only address or represent an extremely small issue or consideration, Thompson said.

Final thoughts

Hennig observed that the customers he believes are “really winning” are those that integrate their IT operations and use the best of both private cloud or their on-premises infrastructures and public cloud. They are making them work harmoniously by adding and extending infrastructure capability where and when necessary and getting the cost and operational benefits and application performance benefits across both platforms for their critical applications, he said.

“So, I would just reinforce that best practice of making this all work together and modernizing your whole organization and following that path,” Hennig said.

Main added that, as organizations look to implement operating models across hybrid multiclouds, it’s essential to always be thinking about flexibility and having the ability to shift at inflection points.

“Always give yourself a way to change course as IT changes,” she said. “That is going to be the key differentiator in your cloud journey. Making sure that – regardless of what the business wants to do, where technology drives you towards and where your people are taking you – it is a flexible journey forward.”