PDF of this tutorialSeveral areas within voice and data networks can benefit from commercial in-memory databases. The following are a few examples:
Voice-over–IP (VoIP) Call Management
IP telephony is gaining increased acceptance as an alternative to traditional circuit-based networks, both in wireline and wireless arenas. To accomplish the goal of ensuring toll-grade, reliable phone service equaling that of circuit-based systems, IP–telephony equipment providers must optimize every part of the network that affects the quality of service (QoS) of a phone call.
The IP call-agent server is a centralized call-processing control center that interfaces with all the customer database and network information required to manage such real-time processes as subscriber service authorization, authentication, address translation, and call management. To ensure reliable service, these systems place a premium on high volumes of database transactions. Furthermore, with their increasing market acceptance, IP call agents must handle unprecedented growth in call traffic.
Commercial in-memory databases can enable the IP call agent to manage time-critical subscriber authorization and network information in real time. Ultimately, they can help boost IP platform throughput and profitability, while ensuring the high level of responsiveness required. IP call agents must quickly process high volumes of subscriber and network information to serve the needs of the caller. In-memory databases can process data with the responsiveness and scalability required. Replication may be used to ensure high availability (see Figure 3).
Figure 3. In-Memory Databases Within an IP Call Agent
Digital Cross-Connect Switching
With the explosion of the Internet and increasing demand for call traffic, networks are growing in number, size, and complexity. As a result, network infrastructure equipment, such as the digital cross connect (DCC), is increasingly critical to ensuring that various networks are instantaneously and effectively interconnected.
The primary function of a cross-connect is to control and maintain the transmission of high-capacity voice and data circuits. With high-speed cross connects, network service providers can quickly adapt to the network demands of a particular region and reroute traffic accordingly. And if a circuit goes out of service, the cross-connect must be able to restore service immediately by rerouting traffic in real time.
In-memory databases enable a cross-connect to perform time-critical network decisions. Managing network status and configuration information so that the cross-connect can perform network routing operations in real time, in-memory databases can improve the overall reliability of the network. In addition, their high performance in data management increases throughput per second, increasing the DCC's bandwidth, capabilities, and scalability.
Figure 4. In-Memory Database Managing Network Decisions in a Digital Cross-Connect
Service and Policy Management
As Internet traffic continues to grow, and the use of advanced network services becomes critical to maintaining a competitive edge, Internet service providers (ISPs) must manage resources and subscribers with increased sophistication and efficiency. Toward that end, service providers are increasingly turning to a service management system (SMS). An SMS lets service providers create and provision IP services—including Internet access, e-mail, VPNs, and LDAP directory services—with advanced policy management functions. The SMS streamlines the management of subscriber provisioning and resource allocation, controlling all relevant network components from a single centralized point in the network.
To deliver subscriber services efficiently, the SMS places extreme performance demands on its data management component. As it performs authorization, authentication, and accounting functions for Internet users, the SMS uses predefined policy rules about the subscriber and the network to control and deliver subscriber services. The SMS must manage such data as session status as well as information about network, subscriber, and policy configurations. Furthermore, it must manage this information in real time during each individual subscriber's session, even as the number of subscribers for some service providers runs into the millions.
In-memory databases offer the real-time responsiveness required for this data-intensive processing, managing the dynamic data required for Internet service provisioning and resource management. In-memory databases deliver the throughput required to incorporate increased intelligence into the SMS while offering the scalability needed to meet the explosive growth of the Internet. In addition, their replication may enable multiple instances of the SMS to be synchronized, ensuring high availability in the event of a network failure.
Figure 5. An In-Memory Database within a Service Management System
SCP Call Processing
INs are specialized networks managing advanced features for mobile and land-line services. With its adherence to industry-standard protocols, IN represents the foundation of an open architecture for delivering communication services. IN platforms continue to demonstrate the required reliability while simultaneously enabling a breadth of new services to gain rapid market acceptance.
The successful adoption of IN services also brings challenges. Along with being called on to accommodate more and more subscribers, these platforms are incorporating increasingly personalized, data-intensive services. To handle this increased capacity and complexity, these platforms are being pushed to their performance limit to retain the required response times.
The SCP is a data-intensive network element that enables the deployment of new personalized services and that is critical to ensuring the responsiveness of the entire IN infrastructure. To process advanced applications such as LNP, number-translation services, VPNs, and prepaid billing, the SCP must search such data as subscriber and service information accordingly. As a result, removing the bottleneck in database transactions allows SCPs to improve their scalability, yielding capacity for more subscribers and new services.
Commercial in-memory databases can help remove the bottleneck in database transactions, enabling the SCP to access time-critical subscriber and feature information quickly enough to process a call in real time. With replication, they can synchronize multiple data stores, ensuring continued data availability in the event of a system failure. By featuring high performance and load-sharing capabilities, in-memory databases ensure that the SCP delivers the high database transaction rates needed to provide service scalability and service reliability.
Figure 6. Replicated In-Memory Databases within the SCP
Call Center
Increasingly, businesses are turning to advanced call-center solutions to improve customer service—a critical effort in terms of remaining competitive. While call-center systems can yield significantly improved customer care, these systems often lack the scalability to maintain effective call routing during high call volume times.
To maximize the potential of call centers for improving customer satisfaction, calls must be routed according to such data as the customer's last five calls, customer's location and products, agent's area of expertise, etc. This information is stored in a database, which is typically the bottleneck of the call center during high-volume times.
This lack of scalability hinders call-center enhancements in two ways. First, they are not able to incorporate additional amounts of customer and agent data that enable calls to be routed with more intelligence. Second, they cannot handle the significantly increased data volumes that arise as organizations seek to integrate Web, e-mail, VoIP, chat, and fax services into their customer-care systems.
In-memory databases remove the bottleneck in the database transactions of a call center, allowing a call center to access and process high volumes of customer and agent data in real time. They can help improve the scalability of call centers so these call centers can handle high call volumes, enabling the integration of more advanced services such as intelligent call routing, automatic number identification (ANI), and intelligent screen transfers (see Figure 7).
Figure 7. In-Memory Databases within Call Centers
As networks become increasingly complex, network management has become critical for providers to maintain reliable service. The network manager is responsible for maintaining network service up time—collecting event information, processing it, communicating it to network operators, and taking corrective action in the event of a system failure.
To be effective, a network manager must check status and configuration information in real time and so ensure that incoming calls and transactions can be properly processed. As a result, improving the performance of database transactions allows a network manager to improve its response time to network events.
Providing microsecond response times, in-memory databases can offer the real-time responsiveness required by network managers. These databases can read and update time-critical status and configuration information, enabling the network manager to acknowledge and respond to critical events in real time and improve overall service reliability. When managing time-critical device-status data, in-memory databases enable real-time response to network events, ensuring that network-processing errors are corrected quickly.
Figure 8. In-Memory Databases within Network Manager
