In-memory computing is increasingly being used in mainstream products and enables extremely fast processing. It gives businesses real-time insights so they can respond immediately and this is why it is so ideal for implementation in hybrid transactional/analytical processing (HTAP).
Table of Contents
What is in-memory computing?
In-memory computing (IMC) has evolved because traditional solutions based on disk storage and relational databases cannot support today’s business intelligence needs for extremely fast computing and data scaling in real-time.
Ram storage processing and parallel distribution are the two main aspects of in-memory computing (IMC). IMC allows for the storage of data in RAM across a cluster of computers. It makes possible the processing of data in parallel and is way faster than the traditional spinning disk. Everything can be placed in an in-memory data grid and distributed across the horizontally scalable architecture. This drastically increases the performance of applications.
In-memory computing is not a new concept but justifying its cost was difficult when RAM was so expensive. The cost of RAM has dropped year by year over the past decades and in-memory computing has become a more affordable option. The value gained from it in terms of performance is making it cost-effective in more and more use cases. The cost of IMC today makes it a practical solution for businesses across all industries.
As IMC distributes processing across a cluster of dedicated servers, it is easy to insert between existing applications and data layers. This can provide great performance and scalability to existing applications. Adding new servers to the cluster allows for easy and cost-effective scaling. With distributed computing comes high availability and super-fast performance, addressing the increasing need for speed.
When building real-time applications, being able to store more data in-memory offers a foundation for improved digital experiences, such as faster and more personalized mobile apps or a richer data set for business intelligence.
What is hybrid transactional/analytical processing (HTAP)?
Hybrid transactional/analytical processing (HTAP) combines transactions and analytics on the same data set. This means organizations can carry out analytics on incoming transactions in real-time.
In the past, transaction processing and analytics would be in the same data set. However, the increasing size of data sets meant that queries would slow down the system and could even bog down or lock up applications.
To process transactions faster, businesses would deploy online transaction processing (OLTP) systems to record and process transactions. Extracting, transforming and loading of data from an OLTP system into an online analytical processing (OLAP) system took place periodically, usually on a daily or weekly basis.
This type of architecture worked well for some time. However, when using separate systems, by the time data arrives in the OLAP system, real-time analytics are impossible. Today’s omnichannel customer experience initiatives require the analysis of huge amounts of data in real-time to drive business decision making and digital transformation became necessary.
Another disadvantage of using separate systems is the necessity of maintaining separate architectures. Hardware and software costs for both systems can mount up and human resources to build and maintain them can also be costly.
How IMC and HTAP work together
With IMC, the whole transactional data set is already in RAM and using in-memory computing platforms makes it possible to run analytics across this data set without impacting transaction processing. Replicating operational data to an OLAP system isn’t necessary and the disk I-O that prevents workloads from happening in real-time is eliminated so it can happen at low latency.
The fact that IMC supports real-time analytics on live transaction data makes it ideal for HTAP. It isn’t possible to perform all data analytics using HTAP but it gives businesses the ability to respond immediately when something happens that affects business operations. The best way to improve customer experience, for example, is to use analytics during an interaction, in real-time.
With no necessity for separate databases, an IMC-powered HTAP system reduces the complexity and costs of data layer architecture. It eliminates duplicate costs as it reduces the technologies in use and uses only one infrastructure.
Some companies use IT support services to help them utilize new technology and give them a competitive edge. Skilled personnel and the ability to use proven open-source IMC platforms can also reduce the costs of implementing IMC technology.
In most cases, the only cost-effective way to get the performance and scale of HTAP architecture is by using an IMC platform. By being able to run analytics over changing data, it is possible to meet the needs of modern workloads. With reduced data and query latency, businesses get predictable performance and horizontal scalability.
Use cases
Industrial IoT: HTAP enables real-time capture of incoming sensor data and making of decisions simultaneously. This offers financial benefits due to higher utilization of assets and predictive maintenance.
Banks and financial services firms: Banks can analyze transactions across many systems to help them identify and prevent fraud. Financial services firms can process countless transactions and analyze risk and capital requirements to meet real-time regulatory reporting requirements.
Online retailers: Retailers can analyze purchases in real-time to update their inventories, adjust item pricing, and optimize their recommendation engines.
Shipping companies: These companies can continually analyze sensor data from delivery vehicles to predict maintenance requirements. This can reduce costs and increase utilization.
Health care providers: They can constantly analyze huge transactional data sets collected from all kinds of sources to respond in real-time to individual crises and examine trends to detect possible disease outbreaks.
A final word
The tremendous growth of data and the importance of being able to make real-time business decisions are driving digital transformation initiatives. Increasing use of the Internet of Things and web-scale applications require an ever-increasing need for scalability and fast performance.
In-memory computing with hybrid transactional/analytical processing can provide a solution that’s worth considering. The competitive need to make real-time decisions by analyzing data is only going to become more critical and IMC-powered HTAP offers a cost-effective path to achieving the necessary speed and scalability.
