How to Modernize Your Legacy Data Warehouse

Tech Insights for Professionals

Tech Insights for ProfessionalsThe latest thought leadership for IT pros

Tuesday, July 30, 2019

What does a modern data warehouse look like, and what features will you need to focus on to create an effective solution?

Article

In today's environment, every firm needs an effective way of storing, processing and analyzing data if it wants to be successful. Over the last few years, the volume and variety of information generated by businesses has grown rapidly, and this is putting huge pressure on legacy systems that are often struggling to keep pace with this new reality.

It's said that as much as 90% of all data currently in existence has been created in the last two years, and around 2.5 quintillion bytes are generated every day. All this needs to be stored somewhere, so the need for effective warehousing solutions has never been more acute.

But many firms will still be relying on legacy systems whose architecture may not have changed in years. And if this is the case, they’re likely to find it difficult to operate effectively in a data-driven world where demand for fast, effective analytics of vast quantities of data is a central part of many enterprise strategies.

If this sounds like your company, then you need to act quickly. The longer businesses delay in upgrading their data warehousing solutions, the further behind the curve they’re likely to fall, as the growth of data volume and variety shows no sign of slowing down any time soon.

Indeed, with more devices and users now online than ever, from smartphones to smart sensors, it’s imperative that steps are taken now. But what exactly should a modern data warehouse look like, and what will you need to do to transform existing solutions so they're fit for today's world?

The need to modernize data warehouses

In order to work effectively, today's end users need fast, reliable access to the right data at exactly the right time. But this can be a problem with legacy systems and those that have expanded in an uncontrolled, ad-hoc manner over the past few years as firms struggle to keep up with their ever-growing data needs. This often means businesses end up with an "accidental architecture" that is more of a tangled web of connections than a single, coherent resource that end-users can connect with.

These sprawling environments will become a particular hindrance as more business units seek to take control of their own data rather than leaving it all in the hands of the IT departments. Today's applications are designed to promote self-service by allowing any users to access key data and analyze it for insight whenever they need it. This means being able to provide individuals with continuous access to the information they need, which is difficult with outdated legacy solutions.

Other common issues with these tools include slower speeds, high costs and a lack of interoperability. Solutions that lack parallel processing will not be able to give users the results they need quickly enough to meet the demands of today's real-time application, while tools that rely on proprietary software will increase overall cost of ownership and make it harder to integrate with other tools.

What defines a 'modern' data warehouse

It's clear that modernization should be a priority for any business that is still reliant on legacy technology. But when does 'modern' actually mean in this context? Before embarking on any such project, it's important to define exactly what this term will cover and what capabilities and features will be deemed essential if the end-product is to meet that definition.

There are several elements that make up a modern data warehouse. For example, it must offer support for all types and levels of user, which will be essential in meeting the demands of business units for easy access to data and promoting self-service solutions.

The ability to support real-time analysis on high-velocity data is another must-have. The use of technology such as a Lambda architecture provides access to batch processing and near real-time processing with a hybrid approach. Additionally, your warehouse should offer support for advanced analytics tools.

Modern data warehouses also need to go beyond traditional relational structures to include support for tools such as Hadoop, data lakes and NoSQL technologies. Indeed, for many applications, creating a data warehouse that can coexist alongside a data lake will be a critical component for many users, as the two technologies each have their own strengths and weaknesses that make them suited to different purposes.

Other features that are defined as part of a modern data warehouse include:

  • Data virtualization in addition to data integration
  • The use of automation tools to improve speed, consistency and flexibility
  • Data cataloging tools to facilitate data search and document business terminology

Updating a legacy data warehouse environment

However, while adding these capabilities will be a major project for businesses, you should be able to introduce features and solutions around existing data warehousing solutions without the need to rebuild from scratch.

When looking to grow and modernize a system, key features that should be considered include partitioning, clustered columnstore indexing, in-memory structures and parallel processing.

As well as growing legacy data warehouse with these features, they should also be extended in order to handle new and emerging data processing solutions. Adding data lakes, Hadoop, in-memory models and NoSQL tools will be essential in building a tool that functions both as a data storage repository and offer analytical solutions.

There are essentially two ways to look at a modern data warehouse when it comes to advanced analytics. The first is a solution where the data warehouse incorporates Hadoop directly, essentially becoming a part of the data analytics platform. The advantages of this include the fact that there are many open-source projects available to assist with this, meaning businesses can utilize distributions such as Hortonworks, Cloudera and MapR. However, it can be challenging to implement.

The alternative is to set up an environment where the data warehouse and Hadoop coexist and complement each other. This is typically an easier objective for businesses to achieve, and it will also augment and add value to existing data warehouses without the need to make fundamental changes to the warehouse itself.

Ultimately, modernizing a data warehouse will provide businesses with a system that is integrated with multiple systems and offer a single source of data, which makes it significantly more valuable than legacy solutions.

For example, having a complete view of a customer in one location lets users view sales activities, any outstanding invoices and any support or help requests that have been logged, helping businesses make much better-informed and personalized decisions, as well as ensuring employees have all the details they need at their fingertips, whenever they need them.

Comments

Join the conversation...

Back To The Top!