If you’re like many business leaders today, you’re thinking of moving away from a cloud-first strategy. You may be doing so for different reasons, such as avoiding higher-than-expected costs, addressing concerns around privacy and compliance or ensuring the flexibility to deploy the right workloads in the right locations.
You may have heard that hybrid infrastructure can help with these and other cloud challenges. If so, you’re likely wondering what the first step in your hybrid infrastructure journey would be.
In our previous blog post, we established that both cloud natives and traditional enterprises are converging on a hybrid infrastructure strategy. We also looked at how colocation data centers provide the foundation for both groups to build the hybrid architecture they need. Working with a top colocation provider can help businesses deploy private clouds in close proximity to their chosen public cloud providers, no matter where they operate. It also gives customers easy access to the low-latency connectivity needed to distribute workloads between private and public infrastructure without compromising performance.
Now, we’ll continue this discussion by examining the steps you can take to get started. Whether you’re considering hybrid infrastructure for your business or simply evaluating what it will take to replatform parts of your environment, this blog post will let you know what to expect.
Developing hybrid infrastructure with colocation services
When you implement colocation as the foundation for your hybrid infrastructure strategy, it’s crucial that you take a methodical approach. Here’s a step-by-step guide to help you navigate the process:
Step 1: Understand your platform usage
First, you must have a comprehensive understanding of how you’re using your current platform. You’ll need this in order to build a hybrid platform that performs comparably. Your analysis should cover several key aspects:
- Geographic distribution: Consider whether your user base is locally concentrated or globally distributed. This will influence decisions about data center locations and the need for edge compute solutions.
- Interaction patterns: Analyze the nature of interactions with your platform. Are they characterized by numerous small requests, or do they involve steady-state operations like data streaming? Understanding these patterns can help you optimize your network configurations and choose the appropriate hardware.
- Latency sensitivity: Assess how critical low latency is for your user experience. Applications requiring real-time responses may benefit from strategically located data centers at the edge.
- Usage patterns: Identify peak usage times and traffic patterns. This information is crucial for capacity planning. It also helps with identifying which workloads are suitable for colocation data centers and which might benefit from cloud elasticity.
Step 2. Analyze your platform architecture
Next, dive deep into how your platform operates to determine your architectural requirements:
- Component interactions: Map out how your different application components interact with each other. This will help identify which parts of your system would benefit most from a colocation environment.
- Infrastructure requirements: List the requirements for each component, including compute, storage and networking needs. This will help you make informed hardware decisions in the colocation facility.
- Data flow analysis: Understand how data moves through your system. Identify bottlenecks or areas where you might improve performance by optimizing your network.
- Regulatory compliance: Consider the regulatory requirements you face across all jurisdictions. These requirements might influence where you can store or process datasets.
Step 3. Evaluate individual workloads and data sources
With an understanding of your platform usage and architecture, you can now assess your individual workloads and data sources.
Identify best-fit colocation candidates
Workloads that are a good fit for moving to colocation include:
- Stable, predictable workloads with consistent resource requirements
- Performance-sensitive applications that benefit from dedicated or specialized hardware
- Data-intensive operations that could leverage the high-bandwidth, low-latency connections offered by some colocation providers
Recognize cloud-optimized workloads
Workloads that are better suited to remain in the cloud include:
- Highly variable workloads that benefit from cloud elasticity, which generally make use of features like Auto Scaling groups in AWS
- Services that leverage cloud native features that can’t be replicated easily in a colocation environment, such as AWS Simple Notification Service (SNS) or Lambda
- Distributed applications that rely on global cloud networking solutions like AWS Shield or CloudFront
While it’s possible to move these workloads, it often requires a lot of effort to refactor them, and that effort must be weighed against any potential gains. Unless there’s a compelling reason to move them, they should remain where they are.
Consider your data management strategy
When implementing hybrid infrastructure, where you host data and how you move it matters:
- Master data: To ensure efficient access and control, your critical data should remain outside the cloud. Keeping your master data copies in a colocation environment ensures that you retain full control over that data. This approach can also provide better performance for data-intensive operations and help avoid vendor lock-in.
- Data location: With your data usage patterns modelled, you can now decide where to locate your data sources. You should consider access requirements, latency, regulatory constraints and data custody.
- Cloud for backup and archival: While primary datasets are best stored in colocation facilities, cloud services are excellent for backup and archival purposes, due to their scalability and geographic reach.
- Data synchronization: For workloads split between colocation and cloud, you need a clear strategy for data synchronization. This might involve real-time replication of critical data or scheduled synchronization for less time-sensitive information.
Assess your in-house capabilities
If you’ve relied heavily on cloud services in the past, you may not have the in-house expertise to move away from the cloud. For example, if you’ve outsourced database management to a cloud service like Amazon Relational Database Service (RDS), you may no longer have database administrators on staff. This means that moving database operations into a colocation facility might prove difficult.
Working with the right partners can help overcome this challenge. Managed service providers and specialist consultants can fill skill gaps, providing the expertise you’re lacking in areas like database management. This can help you take advantage of colocation benefits without having to rebuild entire teams from scratch.
Case studies: Enterprises and cloud natives converge on hybrid
In recent years, we’ve seen traditional enterprises and cloud natives take different paths to end up in the same place: a hybrid infrastructure model. Let’s look at one example from each group.
Target goes from traditional enterprise to hybrid
Any brick-and-mortar retailer that’s still thriving today has undoubtedly overcome fierce competition from online retailers and the limitations of their legacy infrastructure. Target, a U.S. retailer with nearly 2,000 stores, is a powerful example of this.[1] The company now runs a hybrid cloud architecture, and expects to continue doing so for the foreseeable future.[2] However, the path they took to get there wasn’t straightforward.
Target began their cloud journey more than a decade ago. They rely on cloud services to scale capacity as needed. This capability is especially valuable when demand spikes during the holiday shopping season.
The company shifted to various public cloud providers along the way. Each shift required a massive undertaking with significant rewrites. To enable a hybrid multicloud approach without the challenges of lift-and-shift migrations, Target modernized their application stack to help make their workloads fully portable.
Target also experienced public cloud costs that were higher than expected. To optimize cost-efficiency, the company invested in tools to help better understand their service consumption, and to better manage workload placement and data movement in their hybrid environment. As a result, the company slowed their public cloud growth to single digits during 2020-2021, despite experiencing unprecedented digital growth during that time.
Atlassian goes from cloud-first to hybrid
Software providers are ideally positioned to understand the tradeoffs that lead companies to pursue hybrid infrastructure. Not only do they have to manage their internal infrastructure, but they also have a front-row seat to their customers’ considerations about how to use—or not use—the cloud.
Atlassian, the provider of leading collaboration and productivity solutions like Jira and Confluence, was founded in 2002. They’ve been around too long to be truly cloud native, but they were an early adopter: They released the first edition of Jira Software Cloud in 2011. In 2018, they ported Jira Software Cloud to AWS-based microservices, with the goal of maintaining performance and reliability as they grew and adding new product features quickly.[3]
With this and similar product updates, Atlassian made no secret of their intention to eventually move their entire customer base to cloud versions of their products.[4] However, they’ve since softened their cloud-first stance, particularly for large enterprise customers. Their Q4 2024 shareholder letter read:
“Many of these enterprise customers will move to Cloud over a multi-year period, and an increasing number will adopt a hybrid approach of both Data Center and Cloud as they shift their teams and users over time.”[5]
In particular, security and compliance issues might make some customers unsure about adopting cloud-based products. Atlassian is working to address these concerns, but there will always be external factors that determine how much cloud a business is comfortable with.
For instance, many organizations are changing their approach to cloud as they plan their AI strategies. They recognize the need for private AI infrastructure—at least for certain use cases. In an interview with TechTarget, one Atlassian customer said, “I really want to be on the cloud…They are building [a] great ecosystem of connected tools. But if we ever go with [an LLM], we will train one in-house.”[6]
Start your hybrid infrastructure journey today
Traditional enterprises and cloud natives may have had different experiences with cloud in the past, but every business has similar reasons for pursuing hybrid infrastructure today. They all want to optimize cloud costs, ensure the flexibility to move workloads on demand, and keep better control over their data. And many of them recognize that colocation can help them achieve their goals.
With our global portfolio of Equinix IBX® colocation data centers and our market-leading share of low-latency cloud on-ramps, Equinix is well positioned to help our customers successfully implement hybrid infrastructure. To learn more, read our leader’s guide to hybrid infrastructure.
[1] Our Locations, Target.
[2] Hari Govind, Target’s Cloud Journey, Target Tech Blog, October 21, 2022.
[3] Beth Pariseau, Jira update flexes cloud collaboration features, TechTarget, October 18, 2018.
[4] Beth Pariseau, Atlassian ‘cloud-first’ becomes ‘enterprise-first’, TechTarget, October 9, 2024.
[5] Shareholder Letter, Q4 FY24, Atlassian, August 1, 2024.
[6] Beth Pariseau, Atlassian ‘cloud-first’ becomes ‘enterprise-first’, TechTarget, October 9, 2024.
Subscribe to the Equinix Blog