November 5, 2019

By Mathias Golombek, CTO, Exasol

The business benefits of IoT are clear, and nearly infinite. By extracting real-time data from connected endpoints, IoT enables organizations to make more informed and intelligent decisions that can directly impact their bottom line. Too often, though, businesses get swept up by the promise of IoT and focus their resources on gathering IoT-related data just for data’s sake. What makes the IoT so impactful, though, isn’t necessarily all of the data it produces. Rather, it’s the insights that can be derived from that data that’s so valuable.

For IoT, Lightning-Fast In-Memory Technology is Paramount

When troves of real-time data gathered by IoT devices is funneled to a slow analytics platform — as is the case for many organizations — the entire value prop of IoT begins to crumble. Certainly, some businesses may be able to eventually identify some patterns across their incoming IoT data via a slow analytics platform, or even through manual data science efforts. However, today’s issues aren’t necessarily going to be the same as yesterday’s or tomorrow’s, so if organizations want to extract real value from IoT, they need to leverage lightning-fast in-memory analytics platforms.

In-memory refers to using a computer’s random access memory (RAM) as opposed to its hard disk drives or flash memory storage. The technology runs magnitudes faster than a hard disk drive, making it the ideal solution for analyzing large IoT data volumes in real-time. When applied to analytic databases, in-memory capabilities allow for an unlimited number of users to interact with data, while also ensuring the system doesn’t grind to a halt when the most basic of analytics questions are queried.

Best Practices for Acquiring Optimal In-Memory Systems

Still, searching for a new in-memory analytic database system (or even types of databases in general) can be an overwhelming process. A crucial first step is identifying the business applications that need to benefit from performance in order to provide a better service, better products and/or optimize operations. Next, organizations should clearly specify the criteria needed in a database and define a complete set of benchmark requirements. From there, it’s best to establish a solid team from different business units who will conduct the research and buying processes.

Once clear goals and cross-company stakeholders have been identified, organizations should consider the following five factors when researching in-memory analytic databases:

  • General System Architecture: It’s important to remember that every solution is different, and true analytic performance can only be guaranteed through a tight integration of in-memory computing, as opposed to just adding a cache. By taking an integrated in-memory computing approach, database users can run larger and more complex analytic workloads, in addition to using the database for a wider range of use cases.
  • Costs and Scalability: When preparing to implement an in-memory analytic database, always consider the software acquisition or licensing costs, as well as the hardware investment. Additionally, inquire if the database is a scalable massively parallel (MPP) system where additional servers can be easily added. If the system isn’t an MPP, be prepared to incur further costs as requirements and data volumes continue to grow.
  • Integration: A vital step in selecting an appropriate in-memory database is investigating whether the solution is mature enough to handle complex analytic workloads. For example, can the system support commonly-used drivers and interfaces? Does it integrate with the most widely-used Extract, Transform, Load (ETL) and business intelligence (BI) tools? It’s imperative that as analytic ecosystems adapt over time, the in-memory database remains compatible in the future.
  • Vendor Maturity and Customer References: It’s not enough to find a technically solid solution; organizations need to confirm that they can depend on their in-memory database vendor and its customer ecosystem for ongoing support. In addition to asking about the levels of support the vendor offers, consider reaching out to existing customers to discuss the system’s real-world advantages, as well as any shortcomings. If that’s not possible, ask the vendor to supply a variety of customer success stories.
  • Simplicity: As a final step in the research process, determine whether the solution in question is easy to install and operate, or if an army of database administrators is required to tune, design and implement the database and any ETL processes. The more automated the solution, the fewer hurdles organizations will have to incur in order to derive value from its analytics and BI projects.

Success Requires High-Performance, User-Friendly In-Memory Analytic Databases

The whole point of IoT is to be able to make quick, informed decisions based on real-time insights. Organizations can’t afford to be waiting on queries if they want to realize the value of IoT and stay ahead of savvy competitors. Leverage in-memory analytic databases to simplify your existing IT infrastructure and process larger IoT data workloads with far fewer hardware resources. By offloading historically troublesome applications and processes to in-memory systems, organizations can avoid expensive upgrades for legacy systems, traditional databases and hardware appliances — all while maintaining a constant focus on running complex analyses in near real-time to find actionable value within IoT data.