Adopting open architecture for robust data strategy

As the world’s economy grapples with continuous challenges and uncertainty remains the prevailing theme, data emerges as the modern-day fuel, powering enterprises across various sectors. Data and analytics have a pivotal role in helping companies not only manage and alleviate risks but also discover opportunities for adjustment and expansion. Transitioning to data-centric business models is at the heart of the digital transformation wave that continuously impacts every industry throughout 2023 and beyond.

Therefore, the right data storage solution is critical in meeting a business’s requirements. At its core, a data storage system’s role is to securely store valuable data that can be easily accessed and retrieved on demand.

However, as data volume continues to expand and stack up, the need for businesses to increase their storage capacity has grown. The situation gets complicated when data warehouse providers preserve data in their unique, proprietary format. This locks the data into a specific platform, making it a costly challenge to extract when needed by a business. Organisations can quickly fall into a downward spiral while adding more data to their systems given the myriad of data storage options and system configurations – an approach far from efficiency.

Instead, businesses must adopt open-source standards, technology, and formats to allow fast, affordable analytics and flexibility to capitalise on emerging technologies without straining resources.

On the path to open architecture

In the past, businesses leaned heavily on traditional databases or storage facilities to satisfy their Business Intelligence (BI) needs. However, these systems had their fair share of challenges, such as tech interoperability, scalability and costly investments in onsite hardware to maintain structured data in owner-specific formats. It also required a centralised IT and data department for conducting analytical tasks.

Since then, data architecture has experienced dramatic shifts in operations, able to analyse vast datasets in seconds and data quickly moving to cloud storage. This became more appealing for businesses who otherwise would have struggled with physical storage. The transition to cloud warehouses helped elevate scalability and efficiency.

Despite these advancements, certain practices persisted. Data still needed to be imported, uploaded, and replicated into one unique proprietary system linked to a standalone query engine. The deployment of several databases or data warehouses mandated the accumulation of numerous data replicas. Furthermore, companies continued to bear costs for the transportation of data into and out of the proprietary system.

It has all changed with the emergence of open data architecture. As data operates as a standalone layer, it emphasises the distinct delineation between data and compute, where open-source file formats and table formats store the data, and independent, flexible computational engines handle it. As a result, different engines can access identical data within a loosely connected infrastructure. In these arrangements, data is retained as its distinct tier source in open formats within the organisation’s cloud account, allowing customers to access it through a variety of services.

This evolution mirrors the transition from monolithic architectures to microservices in applications. This is especially true in data analytics, where companies migrate from closed proprietary data warehouses and nonstop ETL (Extract, Transform, Load) processes to open data architectures like data lakes and lakehouses.

In addition, separating computing from storage allowed for more efficient operations. Firstly, it lowered the cost of raw storage, practically eliminating them from the IT budgets and leading to significant financial savings. Secondly, the compartmentalisation of computing costs allowed customers to pay only for the resources used during data processing, leading to a reduction in overall expenditures. Lastly, both storage and compute became independently scalable, paving the way for on-demand, adaptable resource allocation that injected flexibility into architecture designs.

Choosing the right data solution

Whilst cloud data warehouses promised organisations scalability and cost-efficiency in data storage, businesses were left tied into the single vendor ecosystem, restricting them from expanding to other invaluable technology solutions. Instead, open data lakes and lakehouses offer a significant advantage for businesses looking to take full control of their data and analytics:

  • Adopting a flexible approach to use various premium services and engines on a company’s data paves the way for utilising diverse technologies such as data-processing tools. With companies having different use cases and requirements, the ability to deploy the most appropriate tool leads to enhanced productivity, specifically for data teams, and a reduction in cloud costs.
  • It’s vital to remember that no single vendor can provide the full range of processing capabilities a company may require. Switching platforms becomes significantly difficult when grappling with a data warehouse containing anywhere between 100,000 or even up to a million tables alongside hundreds of complex ingestion pipelines. However, with data lakehouse, it becomes possible to query the current data with a new system without data migration.
  • Vendor lock-in leads to financial exploitation by vendors, so organisations must avoid it at all costs. However, it’s even more important to be able to merge new technologies, regardless of whether the current vendor continues to be preferred.

    How organisations gather, store, manage, and utilise data will be crucial in shaping the future of nearly every market segment in the upcoming years. Given its vast benefits, organisations must embrace open data architectures if they want to move towards a more flexible, scalable, and insightful future. Only open-source tools can offer the scalability, efficiency and cost-effectiveness that will allow them to stand out from their competitors.

Jonny Dixon is a Senior Product Manager at Dremio. He has worked in the data space for more than the past decade on optimising analytics workloads from both the data and visualisation perspectives. Jonny’s focus is on building ecosystems that build on symbiotic relationships across applications to simplify architectures and make organisations more productive with best-of-breed technologies to democratise data and enable self-service.