Cloud Services / Data / DevOps / Sponsored / Contributed

Modernize Your Cloud Data Warehouse with Real-Time Data

8 Nov 2021 6:30am, by

Josh Treichel
Josh is director of Global Partner Solution Engineering at Confluent, where he leads a team of engineers, consultants and solution architects to drive value with strategic partners. Prior to Confluent, Josh worked for a number of startups building data pipelines, including Rocana and Networked Insights.

The cloud has drastically changed the data analytics space, as organizations have decoupled storage from compute in order to power new analytics, ranging from traditional business intelligence (BI) to machine learning (ML). Gartner projects that 75 percent of all databases will be deployed or migrated to a cloud platform by 2022. As organizations migrate their data from existing on-premises data analytics platforms (Teradata, Cloudera, etc.), they are increasingly moving to cloud-based data warehouses (Snowflake, Databricks, BigQuery, Redshift, Synapse).

However, choosing a cloud-first approach is the easy part. The journey can be long, arduous and expensive, depending on the path you take. To understand why this is, we have to understand how we got to this position in the first place.

Why a Cloud Data Warehouse is the Answer

The data storage problem began with your traditional on-premises data warehouse designed to store and process structured business data, but too expensive to do it in large volumes. These warehouses helped organizations become more data-driven, but had their shortcomings revealed as the volume, velocity and variety of business data increased. In particular, they didn’t separate compute and storage, meaning that as you stored data, it was accompanied by coupled compute resources. Hence, businesses had to make ongoing trade-offs between better data analysis and the high costs that accompanied it.

Data lakes intended to solve this problem by creating a low-cost solution that enabled companies to store, process and analyze vast amounts of data. Often used as a full-fidelity staging area prior to transforming and loading data into a data warehouse, some businesses even tried to use the lake as a replacement for the traditional warehouse. However, data lakes came with their own issues: They required advanced engineering skills to manage and heavy curation efforts. This, along with the tendency to keep all the data led many on-premises data lake initiatives toward the pejorative “data swamp.”

This is where cloud data warehouses are changing the game. These data warehouses separate compute and storage, with the customer only paying for the specific amount of storage and compute they actually use. While that seems like a small change, the ability to store any amount of data and apply compute only when necessary — and only to the data you want to analyze — has dramatically changed this space.

cloud data warehouses are changing the game

To take advantage of these benefits, organizations are increasingly migrating from their traditional on-premises data stores to cloud data warehouses.

Challenges in Moving to Cloud-Based DWs

Challenges in Moving to Cloud-Based DWs

The natural thought at this point is: If cloud-based data warehouses are the answer, why isn’t everyone doing it? As mentioned before, the journey to get there can be long, arduous and expensive. To start with, the sheer volume residing within a typical enterprise has exploded. The average enterprise has more than 400 systems and applications. Simply put, that translates to a lot of data and data pipelines.

Why is this important? Data isn’t just migrated in a one-time transfer of historic data; there are pipelines to be connected, as well as transformations and pre-processing required to ensure that data is usable and production-ready. It’s important to note that while the cloud has changed the economics of warehousing, it’s still not efficient, in cost or speed, to land full-fidelity data into a cloud data warehouse and continuously transform that data in an ELT  (extract, load, and transform) paradigm. Furthermore, these jobs are in many cases mission critical; they cannot suddenly be disrupted and moved.

Finally, many companies want to work across multiple cloud platforms to avoid vendor lock-in and to take advantage of the best-of-breed capabilities suited for their business. This means creating real-time data pipelines across multiple cloud and hybrid environments from across the enterprise.

Move to Cloud Data Warehouses Cost-Effectively

Organizations need a steppingstone as they migrate and modernize their cloud based data warehouses. Specifically, they need a platform for data movement that delivers both familiarity and portability, while helping them drive real-time event streaming and ETL pipelines into DWs across any environment (cloud or on-prem).

Open standards, and in particular open source software (OSS), are important because they are environment-agnostic; they’ll work across any hybrid or multicloud environment, aiding portability and standardization. Further, the familiarity and existing footprint of OSS projects like Apache Kafka in most businesses — used by more than 80% of the Fortune 100 — can make migration and modernization using this technology faster and easier. Finally, an agnostic platform can help standardize pipelines for data landing into any cloud environment. This helps reduce costs and ensures high data quality to modernize your cloud data warehouse.

With Confluent, enterprises can stream data across hybrid and multicloud environments to their cloud data warehouse of choice today, powering real-time analysis while reducing total cost of ownership and time to value. Visit our site to learn more.

Photo by Evgeny Tchebotarev from Pexels.