Consider the scenario of a telecommunications company that uses a real-time billing system to track and charge customers based on mobile data usage. Every time a customer streams video, surfs the Internet, or plays an online game, the company tracks the data transferred and updates the customer’s mobile data balance.
In 2012, this company had enough database capacity to handle and process all customer activity; however, that is changing because mobile data is rising fast. IDC has predicted that mobile Internet usage on smart phones and tablets will surpass internet usage on PCs by 2015. Also entering the mobile data space are cars, household appliances, utility and transportation infrastructure sensors, and other distributed, interconnected systems that make up the Internet of Things. Facing rapidly growing transaction volumes in 2013 and beyond, this company realized its system would soon be completely overwhelmed. How did the company scale its real-time billing system to maintain performance and to take advantage of this huge business opportunity? It used an in-memory data grid.
An in-memory data grid uses intelligent, distributed caching to let applications meet tough requirements of performance, availability, reliability, and elastic scale. It can act as an intermediate layer between the relational database (RDBMS) and the application to hold data that is transient or temporal in nature, or needed for fast access, frequent access, or access across multiple geographies. The RDBMS may subsequently be freed up to store data infrequently accessed or modified.
Today, many organizations struggle to obtain required performance improvements by scaling a traditional software solution consisting of just an application, its local cache, and its on-premise, relational database. In every industry, organizations are discovering that their traditional data tier is too rigid, complex, and expensive to solve challenges similar to those in our example scenario. In-memory data grids have become a popular approach to modernize software and accommodate complex performance requirements. Why? Because, data grids allow enterprises to:
Provide accurate, real-time information. Having real-time access to accurate business information often makes the difference between right and wrong decisions. Data grids move data closer to the application, provide fault tolerance, and enable fast, low latency access to business-critical information.
Meet high uptime and responsiveness expectations. To keep customers loyal and engaged, applications must perform with consistency and deliver seamless service, even during peak activity times and unexpected traffic spikes. A data grid can elastically spin up and down distributed storage nodes helping guarantee required response times.
Process significantly larger transaction volumes. As the amount of data grows, reads and writes to traditional back-end data stores can become a major performance bottleneck for applications. Data grids act as an intermediate layer between relational stores and front ends to meet data-retention requirements and promote extremely fast, scalable performance.
Efficiently integrate with a complex and rigid data-tier. Deploying new applications or updates should be a streamlined, straightforward process. Data grids reduce the overhead and headache that comes with data tier integration, allowing organizations to more quickly go to market.
Interoperate in mixed IT environments. Today’s organizations have a diverse IT environment. Applications and infrastructure may be on-premise, in the cloud, legacy or contemporary. Data grids can serve as a data abstraction layer and offer the flexibility to work in a variety of environments, applications, platforms, and databases.
Simply put, a data grid can act as a supercharger for an application. It boosts application performance to accommodate larger volumes of transactions, spikes in activity, and all at in-memory speeds. Data grids can also be beneficial as part of a big data analytics system, providing functionality such as real-time parallel processing of information, random access, and full text search. And, due to their distributed, fault-tolerant nature, data grids are well-suited to support today’s global, decentralized businesses.
Here are some additional examples of how organizations are using data grids today:
• A global publisher uses a data grid to provide a seamless customer experience across its mobile e-reader and Web applications, achieve reliable performance and scale during peak seasonal demand, and better position itself for long-term business agility.
• A travel organization uses a data grid to reduce the load on the traditional relational database when providing time-sensitive service to its clientele. The data grid holds account and service ticket data in-memory for high availability and accuracy until the service is completed and the ticket closed.
• A worldwide networking company uses a data grid spanning multiple data centers to manage sessions, distribute resources, and handle billing for on-demand streaming video.
Across a spectrum of industries, organizations today face challenges that traditional data tier scaling cannot easily solve. These challenges include providing access to real-time information, handling huge transaction volumes, meeting high uptime expectations, or integrating with a diverse, dynamic IT environment. With in-memory data grids, organizations can achieve flexibility and high performance for their applications and streamline interactions with the data tier—giving them an edge they can use to take advantage of business opportunities and go to market faster.
Christina Wong is a senior product marketing manager at Red Hat, responsible for Red Hat JBoss Data Grid and Red Hat JBoss Portal Platform. She holds an MBA from Babson College and has 10 years of engineering and business experience in high-tech.