Multi-cloud can facilitate innovation and agility—if implemented well. Duan van der Westhuizen, SVP of Marketing & Product at Faction, a cloud data services provider, examines four questions to help define an effective strategy that makes the most out of multi-cloud’s features and benefits while minimizing unintended consequences.
To accelerate innovation, more companies are using services from multiple public cloud providers like AWS, Azure, and Google Cloud. This approach—choosing the right public cloud for each task—is referred to as a “multi-cloud strategy,” and it’s giving medium and large businesses a competitive edge. If you’re wondering why and how to get started with multi-cloud, you’re not alone.
The organizations most successful in implementing multi-cloud have a unified strategy around how teams access resources, governance, and billing. They also remain flexible in how teams adopt clouds and topologies to meet the varying needs of teams, developers, and stakeholders. Before adopting multi-cloud, asking these four questions will help you define an effective strategy.
1. How Is Multi-cloud Different Than a Hybrid Cloud?
The definitions of “multi-cloud” and “hybrid-cloud” can elicit strong opinions from industry leaders. Ask ten people in tech and you’re likely to get ten different answers.
Hybrid cloud blends private and public infrastructure.
To me, a hybrid cloud is about blending private infrastructure (either a private cloud or an on-prem data center dedicated to your applications and environment) with public infrastructure, usually a single public cloud. The hybrid cloud model is attractive because it allows companies to protect specific data and workloads in their private cloud and access public cloud resources when needed.
Some hybrid clouds, like VMware Cloud on AWS, offer operational consistency between the on-prem environment and the public cloud. It offers the ability to use the same VMware-based software-defined data center (SDDC) experience on-prem and in the public cloud, paired with access to Amazon services via an elastic network interface.
Multi-cloud facilitates different workloads running in different clouds.
Multi-cloud—using more than one public cloud for the purposes of gaining advantages unattainable with only a single cloud—is similar, but much more complex. Think about this holistically. It facilitates access of multiple clouds at the same time. As illustrated above, use cases may represent a container workload running in Google; a machine learning (ML), internet of things (IoT) workload running in AWS; and streaming analytics running in Azure. An organization that begins by running machine learning tasks in one cloud would benefit from the holistic management and simultaneous access to features of other clouds as their ML use case evolves into a more advanced solution.
2. What Advantages Does Multi-cloud Offer?
A subset of the many public cloud services currently available.
Agility is a top reason for moving to the cloud. As clouds evolved, they developed services that differentiated from one another; selecting specific services from different clouds is the key to unlocking innovation. (Services grow constantly; those shown above are just a subset of available services.) Reasons behind using different cloud services may include deployment, geographic needs (e.g. availability zones and backup), regulatory restrictions (e.g. from GDPR) about where data resides, and more. The only way to avoid data lock-in (with its associated complexities and egress fees) is to have a toolset that allows you to do the same set of operations across multiple clouds.
Medium to large enterprises may have anywhere from dozens to hundreds of applications at a line-of-business level. Multi-cloud helps developers be agile, improve productivity, and increase innovation by allowing teams to tap into services from across all the clouds. For example, one team that needs to visualize data may use Power BI in Azure, while another team chooses Amazon Redshift as the best way to warehouse the data; one group may build an app on top of AWS Fargate as the best fit for its container strategy, while another team uses Azure HDInsight to process data.
3. Which Workloads Should Run in Which Cloud?
Each organization must consider which clouds are best suited to its needs.
Each organization will have a somewhat different answer here. Broadly, the more critical the workload and the more it is changing, the more you can leverage multi-cloud to provide resiliency, innovation, and security. Whether your customer is internal or external, make careful decisions about your goals. Triage to find easy things that deliver significant value for end users. Consider not just everyday workloads, but disaster recovery (DR) and back-up recovery use cases, as well.
Evaluate how quickly you’re innovating; your staff’s level of expertise with different clouds; security; data sovereignty and auditability; and locality. Is the data center where you need it to be? Is your cloud instance as close to your customers as it needs to be? As more work happens around the edges, location is likely to become a bigger concern.
Today we’re hearing about researchers using more than 30,000 GPUs for COVID-19 analysis. Even if you’re working at a much smaller scale, you may find great benefits if you can leverage spot and low-priority instances. The availability and cost, though, can vary significantly across cloud providers.
4. What Common Mistakes Should I Avoid When Adopting Multi-cloud?
Simplifying multi-cloud adoption relies on having a common governance model, considering data availability, looking for interoperability with existing processes and investments, and finding ways to start simply. Common blunders include:
- Adopting multi-cloud for the wrong reasons: Be deliberate in your adoption of additional clouds to ensure the value add isn’t canceled out by management complexity. Use best-of-the-breed services where they make sense.
- Insufficient governance: Without appropriate governance to meet the needs of security and compliance teams, multi-cloud shadow IT can result as DevOps teams with different perspectives use different tools from different clouds.
- Assuming a lack of change: Recognize that change is the new normal. You’ll have different needs—requiring a different cloud topology and different cloud services—tomorrow. Even if you only have a single cloud platform strategy today, you can prepare for that multi-cloud eventuality; centralized storage that makes your data available across all clouds at the same time helps you be proactive.
- Data sprawl and data gravity challenges: Placing data and compute resources centrally as an enabler to public cloud, you can get 2–10x as much value from that cloud transformation as you might otherwise. Bringing everything into one central location can help you see where you’re storing and paying for redundant data.
- Surprise egress fees: Pulling data out, even over direct connect, can be technologically challenging and colossally expensive, depending on your use case (e.g. active v/s never accessed).
Future-Proof Your Multi-Cloud Strategy
The best way your organization can achieve its multi-cloud goal? Have a strategy. Very commonly, organizations stumble into a process. If you went to the public cloud in the first place for innovation, ask yourself why you should lock your intellectual property (IP) into a single cloud or settle for only 30–50% of the available agility by locking yourself out of the other clouds? Appropriate considerations today can help future-proof your organization’s cloud strategy as you continue to innovate and grow.
To access the original article, visit it.toolbox.com