Where In-Memory Computing is Going in 2021

The adoption of in-memory computing platforms has steadily increased over the past decade, growing in popularity as tra­ditional disk-based databases could no lon­ger provide the performance and scalability businesses needed to collect and analyze soaring volumes of business data. In-mem­ory computing platforms, built using a clus­ter of servers, pool the available RAM and CPU power, enabling data and processing to be distributed across the cluster for blazing fast performance and massive scalability to enable real-time business processes.

Some industries, such as financial services, have embraced in-memory computing for decades. Others, such as food delivery, e-commerce, logistics, and remote access and collaboration services, did so throughout 2020 as companies faced COVID-19-fueled surges in website visitors, delivery requests, sales transactions, video streaming and more. In-memory computing platforms enabled these businesses to rapidly scale to support these dramatic usage increases.

A surge in in-memory computing adop­tion by businesses for a growing number of use cases has led to an increase in in-mem­ory computing practitioners. This in turn has led to greater adoption of in-memory computing technology and the acceleration of innovation for in-memory computing platforms overall.

As we continue through 2021 and in-memory computing platforms mature, we will see the number of industries and companies adopting these solutions con­tinue to grow. This trend will last far beyond the effects of the pandemic as companies shift their focus to meeting new strategic and competitive demands driven by digital transformation initia­tives. Increasing adoption of in-memory computing will also be accompanied by the following key trends powered by the technology’s real-time capabilities and massive scalability.

Rapidly Improving Application Perfor­mance and Scalability

First, watch for in-memory data grids (IMDGs) to become the solution of choice for more and more businesses that need an easily implementable approach to main­taining or improving performance while increasing application throughput.

Because IMDGs cache application data in RAM and apply massively parallel pro­cessing (MPP) across a distributed cluster of server nodes, they provide a simple and cost-effective path to dramatically improv­ing application performance. An IMDG is a component of an in-memory computing platform. An IMDG can be inserted between an existing application and disk-based data­base without major changes to either layer. The distributed architecture also enables rapidly scaleout of applications by allowing the RAM and compute power of the cluster to be increased simply by adding new server nodes. The platform automatically recog­nizes new nodes added to the cluster and redistributes data and compute across them.

Digital Integration Hubs

The next trend to watch is the rise of the IMDG-powered digital integration hub (DIH). Digital integration hubs are rapidly increasing in popularity because of their abil­ity to power one of the hallmarks of digital transformations: real-time business processes that require access to multiple data sources to create 360-degree views of customers and the business. The DIH leverages an IMDG as a high-performance data access layer for aggre­gating a subset of data from multiple source systems. This data can then be simultane­ously accessed by any number of business applications through a variety of common APIs such as SQL or REST at in-memory speeds. Synchronization between the IMDG and data sources which can include relational and NoSQL databases, data warehouses, data lakes, cloud-based applications, and stream­ing data is continuous to ensure the data in the in-memory cache is always current as changes are made to the underlying data­stores. This makes the DIH ideal for launch­ing a variety of critical real-time initiatives, from recommendation engines to automated business decision making to real-time upsell and cross-sell offers.


The last big in-memory computing-pow­ered trend to watch in 2021 is the accelerat­ing adoption of hybrid transactional/analyt­ical processing (HTAP).

HTAP enables businesses to perform real-time analytics directly on their transactional or operational data, eliminating the need for a time-consuming ETL: process to move operational data to a separate datastore. In this way, HTAP enables real-time digital busi­ness across a range of data-intensive indus­tries, from financial services and e-commerce to healthcare, transportation, and logistics.

Increasing HTAP adoption is also tied to soaring demand to roll out applications that can update their machine learning models in real-time. Providing what Gartner calls “in-process HTAP,” an in-memory comput­ing-powered data grid can enable a contin­uous learning infrastructure in which the machine learning (ML) model is updated in real time based on new operational data flowing into the system.

Driving Real-Time Processes

In-memory computing has solidified its place as the most flexible and cost-effective strategy for achieving achieving real-time business processes using all of the relevant data in an organization’s systems As a result, we can expect to see in-memory computing adoption continue to accelerate for years to come as companies rely on it to achieve ever-greater levels of application speed and scal­ability as well as real-time data access across all datastores.