Understanding the SBRC Network Architecture
This chapter provides a reference to a highly reliable network architecture for SBRC. It explains some existing error-handling techniques in SBRC which can be used to improve the overall performance and reliability of the system of systems.
SBRC cluster solution is architected based on principles which permit upwards of “5 9s” (99.999 percent) system uptime as a whole, outside of maintenance windows, making the system highly resilient to failures.
Redundancy is offered at multiple levels in SBRC:
Within the RADIUS transmission structure, built on UDP, NAS devices at the application-level can retransmit to different devices after several attempts to the same device have failed before timing out the user.
Within authentication, accounting, and proxy functionalities, multiple back ends or downstream RADIUS servers are attempted with appropriate retry semantics.
Within the SSR, MySQL NDB is resilient to system outages of a single component (a machine or a network component) and some multi-component failures (for example, cluster splits).
At the networking layer, utilizing techniques such as IP multipathing (IPMP), failures in hardware components such as network adapters, cabling, and switch outages, can be managed.
At the machine layer, using RAID-5 or external storage with RAID mitigates the risk of disk failures. The implementation of Oracle’s advanced system maintenance in terms of hot-swappable drives and power supplies limits the risk of total subsystem unavailability. The regular use of ECC (Error Correcting) memory helps to avoid memory corruption.
This chapter explains: