Skip to main content

Choosing among different cloud data storage solutions is like building your dream kitchen. Identifying the best option requires weighing a host of factors, and unless you know what's most important to you, it's impossible to make the right choice.

You envision a culinary haven with top-of-the-line appliances seamlessly integrated for maximum efficiency. But instead of a cohesive package, you're presented with a jumble of parts from various manufacturers – a Viking range, a budget-friendly countertop, and a mismatched refrigerator.

The result? A kitchen that functions but lacks the harmony and efficiency you crave.

That's what choosing cloud data storage solutions can feel like. A dazzling array of options exists – scalability, storage costs, query costs, query performance – each boasting unique strengths. But without careful consideration, you could end up with a hodgepodge of services that don't quite work together, hindering your team's productivity and jeopardizing data security.

The perfect solution is one that empowers your team and keeps your data safe and sound.

This article is your blueprint for a well-designed cloud storage solution. I’ll walk you through the questions to ask when comparing cloud storage options and then explain which types of workloads align best with which storage offerings.

Cloud Data Storage Options: The Basics

Before diving into the nuances of choosing a cloud storage option, let's discuss the types of cloud storage available today. The main options include:

  • Databases: Databases, which store information using rigid structures, are the most traditional approach to storing large volumes of data at scale.
  • Data lakes: Data lakes are centralized repositories for data. Unlike databases, they can accommodate both structured and unstructured data.
  • Data warehouses: Data warehouses are like databases in the sense that they are designed for storing structured data. But unlike databases, warehouses are typically optimized for specific types of queries. They can also serve as central repositories for multiple types of data, warehouses most databases support a single type of data.
  • Data lakehouses: Lakehouses, which can house both structured and unstructured data, combine the flexibility of data lakes with the query optimizations of data lakes.

There's a lot more to say about cloud-based data storage options and how they differ from each other. But the point here is that there are many ways to store data in the cloud, each with different tradeoffs in areas like flexibility, scalability, and performance. 

Discover how to deliver better software and systems in rapidly scaling environments.

Discover how to deliver better software and systems in rapidly scaling environments.

  • By submitting this form you agree to receive our newsletter and occasional emails related to the CTO. You can unsubscribe at anytime. For more details, review our Privacy Policy. We're protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
  • This field is for validation purposes and should be left unchanged.

Understanding Your Cloud Storage Needs

With so many choices, including traditional cloud storage and even hybrid cloud storage options that combine on-premises storage with cloud resources, which type of cloud storage option is best for a given workload? To answer that question, start by determining what the purpose of the workload is, and what it requires to excel at that purpose.

For example, a workload that involves real-time data analytics is likely to require lower latency than one that processes data in batches. In that case, you'd want to prioritize cloud storage that enables fast queries. Similarly, if the volume of data that you're working with is fixed or predictable – as it might be if you're dealing with application logs that grow at a consistent rate, for example – then scalability is less important in a storage solution.

Another key factor to consider is which ancillary purposes a workload might serve. Sometimes, you need to use one workload for multiple tasks. For instance, imagine a dataset composed of customer information. The data's primary purpose is to support an e-commerce app that needs to look up customers when processing purchases. In addition, your marketing team periodically runs queries against the data to identify trends about your business's customers.

When selecting a cloud data storage solution for the data, you'd want to ensure it excels at the primary use case you need to support while still supporting the ancillary one. Going further, you'll need to determine how much priority to give to each of the purposes: Do you want to optimize the e-commerce app's performance above all else, for example, or can you accept some performance sacrifices for that use to improve query performance when the marketers analyze your data?

Aligning Data Storage Options

To gain a more specific sense of which cloud storage options are best for which needs, here's a look at the areas where different storage solutions excel. These are generalizations, and your mileage may vary, but think of them as a starting point for deciding which type of cloud storage is best for your business's needs.

Multi-purpose data

If your data needs to do many things and they all have approximately the same priority level, a data lake is probably best. Data lakes are the most flexible and open-ended way to store data. The tradeoff is lower performance for some use cases. However, that might be an acceptable sacrifice if a data lake allows you to support multiple data needs using a single repository.

Transactional data

If your data is transactional in nature (meaning the data is generated from transactions, such as user interactions with software applications), your main cloud storage options are databases and data warehouses.

Databases are typically the better option because they deliver high levels of query performance while also being relatively simple to implement and maintain.

AI data

Data used for AI workloads comes in two main flavors: streaming data that drives real-time analytics and batch data dedicated to purposes like model training.

When dealing with streaming data for AI workloads, you'll want a low-latency environment to ensure that you can move and process data in as close to real time as possible. Data warehouses typically aren't ideal for this purpose because they have more overhead. Instead, choose a high-performance database, such as PostgreSQL (or a cloud-based managed version of PostgreSQL, like AlloyDB).

For batch data, however, latency is usually not a priority, but scale is because you often need to work with very high volumes of data. Data warehouses are a good solution in this context because they can handle virtually unlimited amounts of data.

Modern Cloud Storage Unifies Your Information

At this point, you might wonder: What if my data needs are too varied and unpredictable to commit to a single cloud storage option?

The good news is that modern cloud storage tools and services have evolved to make it easy to interact with data spread across multiple storage options efficiently and coherently.

To take Google Cloud Platform (GCP) as an example, if you want to query data on GCP, it doesn't matter which specific storage services host your data. You can use GCP BigQuery to analyze data stored in virtually any of the databases, warehouses, data lakes, or other storage options that GCP supports.

There may be some performance implications; for instance, BigQuery queries run faster when your data lives natively inside BigQuery rather than in a separate part of GCP. Therefore, it's still important to optimize your storage choices based on the intended purposes of your data workloads.

But if you don't make the ideal choices in every instance, don't sweat it. Today's cloud storage services and tools are sufficiently integrated such that you can usually do what you need to do regardless of exactly where your data lives (provided it all lives on the same cloud platform, at least – querying data across multiple clouds is a more complex challenge that's beyond the scope of this article).

Getting the Most From Cloud Data Storage

The many cloud storage options available today mean that businesses have many decisions to make when storing data in the cloud. The best way to approach those decisions is to think in terms of what you need your data to do—and if you need it to do more than one thing, decide which use cases are more important than others.

And if you're struggling to make a choice, remember that integrations between cloud storage services give you a lot of flexibility to achieve your goals regardless of where you store your data – although that's not a reason to avoid trying to select the optimal type of storage from the start.

Subscribe to The CTO Club's newsletter for more cloud storage insights and the latest insights from CTOs.

By Daniel Zagales

Daniel Zagales is the vice president of data engineering at 66degrees. He is an expert at designing, implementing and scaling analytic solutions and working in fast-paced agile environments, driving change, growing adoption and empowering the enterprise to make informed decisions.