Moving to Multi-Cloud? Start with the Architecture.
As VMware CTO, I have the opportunity to meet with customers around the globe and across a variety of industries discussing the challenges they face in organizing their cloud architectures. Based on these discussions and many years experience as a Cloud Architect, I would like to share four key imperatives that can help with the success of your multi-cloud project.
1. Define the Basic Building Blocks
The best way to build a cloud from all the technologies and products available from multiple vendors is to start with a comprehensive top-down cloud architecture. Resist the temptation to skip this phase in the hope that everything will work out. Moving right into assembling the pieces without defining the architecture is unlikely to deliver maximum business impact and can have financial repercussions in the end.
A good architecture is not merely an on-paper exercise. It is a comprehensive architecture containing the nuts-and-bolts details of all use cases and key design considerations, along with implementation guidance on each product and technology, including documentation of the associated dependencies.
For example, it will be difficult to build a horizontally scaling cloud that takes advantage of software-defined networking without first redesigning your traditional, multi-tier network. Only by flattening out the design of your network can you change the operating model for networking services.
Looking at the focus areas in Figure 1, once you have defined your architecture and layered it in a software-defined infrastructure, you are ready to start automating core capabilities. Having automation above the software-defined infrastructure layer allows you to build aggregate services such as IaaS (Infrastructure as a Service), CaaS (Containers as a Service), DaaS (Desktop as a Service) and FaaS (Functions as a Service).
Fig 1. Fundamental Building Blocks and Cloud Principles
2. Embrace the Strategic Cloud Principles
There are four cloud principles that act as strategic levers: cloud economics, cloud scale, cloud speed, and cloud security, specifically Day 2 security.
To successfully leverage these principles, think in terms of the workloads you want to run in the cloud. Often, IT teams are unaware of exactly how their supported applications consume the underlying infrastructure. It is easy to throw hardware at every problem to the point where many applications are running at 30% or lower utilization. Instead, I recommend designing an architecture for the workloads you actually have. Your subsequent decisions then become a technical or business choice.
This begins with questions like “what level of cloud scale do I need for the applications I actually have?” Since there is always a finite limit to your data center, design in a way that provides extra capacity when you need it. You might burst out to AWS or Azure, but you will still need to take your resiliency models, security models, and DR models to the cloud as well.
Cloud speed is a key goal of application development teams seeking always-faster delivery times. Tension tends to form here, as it is common for IT’s controls and processes to add significant time to service delivery. An example is firewall rules: if you provision a VM in 20 minutes but it takes two weeks to get IP addresses and DNS services provisioned, then VM delivery to the app teams takes 2 weeks, not 20 minutes. Service delivery is all about getting the right balance between speed, scale, and economics, while ensuring that necessary security undergirds the entire design.
3. Move Up the Business Value Chain
How will your cloud strategies drive value for the business? In discussions with customers, I have found that infrastructure costs represent, on average, about 20-30% of the costs associated with an application. What ties up the majority of the costs are non-infrastructure related activities like application development, delivery, support, and lifecycle management. Even if you can reduce infrastructure costs, you still have not significantly addressed the overall cost envelope.
So how do you start to affect that cost majority? Figure 2 shows a roadmap of what to think about. Once you have a solid foundation, you then have to address PaaS, CaaS and FaaS capabilities. You also need to address the development of DevOps models that improve Continuous Improvement/Delivery processes. These are the investments that really help make app dev teams faster and more efficient.
Fig 2. Building Business Value Step-by-Step
Recently, a customer told me he had spent millions of dollars with VMware but struggled to explain the benefits to his CEO. I recommended looking at the strategic levers of cloud speed, cloud scale and cloud economics. The customer was successfully balancing these but was still on the IaaS side of the continuum. What he needed was to keep driving towards additional business value, because that is where his CEO, and probably yours too, is looking.
I asked key questions like “How are you leveraging your data?”, “What is your data costing you?” and “What’s the cost per gigabyte and how fast is that cost growing?.” Their fiber channel costs were $1/GB per month and growing 30% a year. I recommended getting to software-defined storage ASAP to get costs down to ~$.10/GB, before storage ate the entire IT budget. This customer also needed to explore ways to monetize the data they generated so it added to the company’s revenue stream and was not merely an ever-growing cost.
At the right of Figure 2 are specific use efficiencies where Build vs Buy scenarios require careful consideration. Finally, you have Business Technology Alignment where you are seeking to align multiple technology services across multiple business units.
4. Have the Right Conversations with Key Stakeholders
Addressing the full scope of this continuum is how you begin to address the 75% of the cost equation – reducing app development costs, increasing the velocity of app delivery, and enabling the enterprise to leverage data to generate additional value from the apps that get built. But how often do we start conversations on the right side of the continuum? Of the 200+ companies I have visited over the past year, most are still focusing their conversations on the left side of the continuum.
To have a real impact on the business, make sure your conversations span the full spectrum. Otherwise, it will be a struggle explain how IT is creating value for the business. It’s important to have these discussions upfront and early. It is equally as important to have an architecture that looks out across this entire continuum and helps you make informed, long-term decisions that drive value for your business.
To recap, following a few key steps can increase the success of your cloud project.
- Define a realistic architecture that starts with basic building blocks addressing the needs of existing applications in your portfolio and apps you expect to deliver in the future.
- When considering your cloud-use scenarios, consciously architect the trade-offs across the four cloud levers of speed, cost, scale and security.
- Always strive to move up the business value chain. This means broadening your focus beyond just infrastructure to understand the core processes surrounding applications.
- Remember: people are your top priority. Cloud initiatives touch many teams, so actively manage the change needed in culture, core processes and technology to be successful.
Ultimately, architects and senior IT leaders must take the lead in developing an architecture that meets the needs of the business. But the good news is, you don’t have to go it alone. Reach out to VMware or another trusted partner to help you and your team be successful in architecting a multi-cloud solution that drives value across the full scope of the business value continuum.
VMware OCTO Application Platforms Position Paper