Break the Bonds of Data Gravity With a True Multi-Cloud Strategy

A collection of data silos in multiple clouds is not the same as a multi-cloud strategy. Instead, a shared data repository makes data simultaneously available across clouds, preventing the need to store multiple copies of the same data, explains John Drake, VP, Strategic Alliances, Faction.

By most accounts, IT innovation is happening at a record pace. The pandemic drove cloud usage and digital transformation. Digitization leaped ahead by 3–4 years, as estimated by McKinseyOpens a new window , with many investments made for accelerated migrations to the cloud.

But while the usage of multiple clouds is going up (and often into silos), data gravity is bringing it down. Data gravity’s negatives may include lock-in with a single cloud provider, stranded data, and high bills. The alternative, a shared data repository, makes the same data simultaneously available for use by multiple cloud services across clouds — preventing the need to store multiple (and possibly out-of-sync) copies of the same data. 

It is worth taking a closer look at the effects of data gravity and how emerging approaches are working to mitigate them. 

Avoid Siloed Cloud Data 

A collection of data silos in multiple clouds is not the same as a multi-cloud strategy. Very commonly, though, organizations end up with a disjointed cloud data implementation because they did not address the need for a coherent approach of integrated cloud initiatives. When workloads are isolated in silos, they hinder effective data management and usage.

The fragmentation caused by silos can lead to wasted resources, “tunnel vision,” and “information bottlenecks,” as described in The Silo Effect: The Peril of Expertise and the Promise of Breaking Down BarriersOpens a new window by Gillian Tett. These risks are undoubtedly present when using multiple clouds without a clearly defined and streamlined strategy. 

In 2021, which IDC expectsOpens a new window “to be the year of multi-cloud,” an effective multi-cloud strategy requires more than relying on a combination of multiple clouds. Multi-cloud requires a common data repository as part of a strategic approach to integrating across an organization’s chosen clouds and provides tangible benefits to your cloud data storage initiatives. 

Learn More: 6 Ways to Leverage Document Automation in the Manufacturing and Distribution Industry

The Problem of a Disjointed Approach

As businesses move to multiple clouds, the process often goes something like this: 

• Take an application. Put its data in public cloud #1, say, Amazon Web Services (AWS). Add AWS compute, network, and storage. 
• Take another application. This time, put it in a second public cloud, say, Azure, with Azure compute, network, and storage.
• Repeat and create a new silo in another cloud.

Initially, this approach has its purpose. The rapidly growing number of public cloud services allows users to select from hundreds of offerings. A user chooses the specific cloud service that best meets their immediate goal. This could be creating high availability, meeting a global regional availability need, desire to use a cloud’s hardware and software offerings, or delivering a particular cost benefit. They then move the workload to the desired cloud accordingly. 

Workloads that rely on large, file-based data sets and artificial intelligence and machine learning (AI/ML) demand significant storage and compute resources. Once the data is in a particular cloud, data gravity — the consequence of attracting and using additional applications and data — is so strong that it becomes very difficult (or even prohibitive) to move data out of that particular cloud, effectively trapping that data in a silo. 

This challenge is cumbersome enough when you are handling 100 terabytes (TB) of storage but becomes extremely challenging with 20-30 petabytes (PB). The amount of structured and unstructured data is easily in the PB-zone for the compute-intensive high-performance workload projects used by various industries, including predictive financial analytics, autonomous vehicles, and genetic sequencing.

Expenses rapidly escalate when trying to move a large data set out of its current silo. Egress fees (which vary by cloud and are determined by the gigabytes of data being transferred out of the cloud) are incurred when you choose to repatriate data to process it. Instead of moving data out of a cloud, you could make a second copy of the data, but that approach doubles the time, complexity, and expense of data management. Multiple IPs, multiple volumes, and complex management with inflated costs and sync issues are just some of the additional challenges created by duplicated sets of data that are each accessible only by a single cloud. 

These complications limit the opportunity to use your data in another cloud as you move forward, preventing you from taking full advantage of cloud service offerings and the innovative opportunities they provide. To simultaneously enable the use of different cloud services — a true multi-cloud strategy — while minimizing overall costs and optimizing performance and availability for a particular workload, a common data repository is necessary. 

Learn More: Transforming Weather Forecasting Using AI/ML-Powered Analytics

Access Your Data Without Moving It

To harness the power of your data and future-proof your organization, it is important not to paint yourself into a corner. By using a single copy of data that is accessible across multiple public clouds, you can create a flat network that can connect to multiple public clouds and your on-prem environment. 

Determine the best place for your workload to run and place it on any cloud without moving or replicating data. This multi-cloud infrastructure strategy allows business users and developers to select their preferred cloud services, optimize each workload’s performance and availability, minimize overall costs, and move beyond the siloed verticals.

Creating a single common data management plane, separate from the application management layer:

• Supports flexibility and fluidity: Turn on or off access to where you want each application to run. This provides fluidity between clouds and instance types, eliminating the need to move data when accessing best-in-class cloud data services from each cloud. The ability to activate a workload and access the data means that you can migrate more workloads to the cloud, scaling to take advantage of available services. As the popularity of container-based applications and Kubernetes grows, this same approach allows for fluidity with containerized environments.
• Improves data analytics: Maximize the use of data analytics platforms (such as Azure DatabricksOpens a new window ) and optimized instances (as provided by YellowbrickOpens a new window Data) to speed queries on PB of data and unlock the value of your data. 
• Strengthens data sovereignty and security: Know exactly where your data is and maintain control of it when you host your data in a central data lake that you can use with any cloud provider. This helps you comply with varying regulatory and compliance requirements.
• Conquers data gravity: Data gravity leads to lock-in with a single cloud provider, stranded data, and high bills. With a single copy of data accessible from any cloud provider, you will not get weighed down by data gravity. 
• Provides tangible budget acquisition: Lower storage costs by 89% when using a single data set instead of native file storage with the same data set on multiple clouds (Enterprise Storage Group, February 2021 paperOpens a new window ). These significant savings help optimize your overall cloud spend and free up the budget for other strategic initiatives.

A verticalized approach to managing your data in the public cloud creates unnecessary segmentation and waste. Move toward a true multi-cloud data services strategy and away from silos. 

Did you enjoy reading this article? Let us know your thoughts in the comment section below or on LinkedInOpens a new window , TwitterOpens a new window , or FacebookOpens a new window . We would love to hear from you!