Article

4 tips for building a successful multi-cloud strategy

Public clouds are here to stay. You know this. It’s only a matter of time before all of your infrastructure and services are in the cloud, so knowing how to build a successful hybrid, multi-cloud strategy is essential. Here are four pieces of advice to help you navigate the multi-faceted waters of cloud like a pro.

Watch out for data silos

Regardless of what storage choices you make, it’s important to keep an eye out for data silos in your organization. They come in many forms, and they should be taken seriously, as they can really hamper the productivity of your organization. 

Common forms of data silos include standalone relational databases, block storage such as persistent disks and data warehouses holding semi-structured data. 

You may have the same data stored in multiple locations, which means you have no single source of truth. You also might end up updating one storage location frequently but not updating another.

There are two approaches to solving data silos: integration or data lakes.

 

Integration

If you want to let your silos exist, you can connect up the different bits of data using ETL or extract transform load pipelines. This allows you to bring important bits of data together for visualization or analytics. 

 

Data lakes

Azure, AWS and other cloud providers have their own data lake solutions you can use. Data lakes allow you to have a single source of truth, and are great if you’re going with a cloud-first approach where you’re planning to do a one-off migration to the cloud. They hold raw data in all formats—structured, unstructured and semi-structured data—and they're usually built around blob storage such as S3 buckets, Azure blob storage or Google cloud storage buckets. 

Avoid vendor lock-in

Here are two rules of thumb for avoiding vendor lock-in on the cloud:

Avoid past offerings or platform-as-a-service offerings for compute. 

Hosted applications will offer you automatic load balancing, version management, traffic splitting  and more—which is incredibly convenient—but if you want to move your app to another platform it may involve a full record. Instead, set up your application using Kubernetes containers. This will    make your applications more portable. 

Avoid platform-specific solutions for ML and AI applications.

Use generic data formats and RDBMS engines where possible

Be a satisficer, not a maximizer

A satisficer is someone who tries to find a solution that checks all of the boxes, then moves on and tries to make the best of the situation. A maximizer, on the other hand, will try to find the best possible solution out there and won’t stop. 

Many studies have shown that satisficers tend to be happier and more successful in business, which is valuable to keep in mind when making technology choices. Whatever your use case, there’ll only be little differences between big flagship products, and each will have their own areas of strength. Trying to squeeze out every last drop of performance optimization will likely lead to platform-specific changes, which could create vendor lock-in.

Sweat the small stuff

Security issues and data breaches occur when little details are overlooked. Cloud-based systems are inherently distributed and replicated horizontally. Scale systems such as these are more complex and need more coordination and integration. Each point of coordination and integration is a potential point of failure, which is why it’s so important to be conscious of the small things. 

The costs of failure are very high. Even if an organization recovers from a data breach, the career of its CIO can be ruined. To stay on top of arcane little threats that explode at random, CIOs should talk to analysts, their support team and engineers often. Once you hear enough stories across the board, you’ll start seeing patterns emerge, which may be obvious to you but less obvious to them. Listening a lot and keeping your ear to the ground is an incredibly valuable strategy for protecting your organization.

When it comes to building a successful multi-cloud strategy, progress isn’t always linear. You may need to experiment for a while before you figure out a really solid system for your organization’s needs. But if you stay focused on your goals, the effort will definitely be worth it.


Working on your cloud strategy? Check out Janani Ravi’s “https://www.pluralsight.com/resource-center/webinars/building-multi-cloud-strategy” webinar on-demand now for free.