CS459/559 - Network Security - Chapter 14: Redundancy

1. Reliability, Availability, and Serviceability
   • Reliability (minimize system failures and ensure data integrity): MTBF(Mean Time Between Failure)
   • Availability (accessibility of the system's data and applications): % of total uptime
   • Serviceability (need for short downtime in event of a failure): MTTR (Mean Time To Repair)
   Sun Microsystems' SunTrust
   • Gigaplane-XB Crossbar Interconnect - provides communications between processors, memory and I/O
   • Dynamic Reconfiguration - enables physical/logical restructuring of the server while the system is active
   • Integrated System Service Processor - central console that monitors the system and enables the operator to carry out management and maintenance tasks.
2. Redundant Cabling - link redundancy, meshed network; automatic switching
3. Redundant Servers - BDC
4. Redundant Storage
   • Disk Duplexing - data is automatically written to two or more disks and sent through separate channels simultaneously
   • Disk Mirroring - data is automatically written to two or more disks one after another through a single channel (cannot use IDE drives - if master crashes, cannot use slave drive)
   • Disk Striping - combines partitions on separate hard disks into a single volume (with parity -distributes parity information across partitions)
   • RAID (redundant array of inexpensive disks)
     • Level 0 - data striping writes data a block at a time across each drive (no fault tolerance)
     • Level 1 - disk mirroring and duplexing
     • Level 2 - data striping and bit interleaving (slow and unreliable)
     • Level 3 - Level 2 + parity checking (uses separate parity drive)
     • Level 4 - Level 3 with block writing instead of bits
     • Level 5 - Level 4 with parity / checksum data distributed across regular disks (considered most reliable)
     examples
     • HP AUTORAID - continually optimizes for cost and performance
     • IBM RAMAC SNAPSHOT - provides a virtual second view of data in near real time
     • NT
5. Redundancy in High Speed Networking
   • Fast Ethernet - can have redundant cabling links requiring special transceivers
   • FDDI - uses dual counter-rotating fiber ring topology to provide fault tolerance (concentrator has two buses to accommodate two FDDI backbone rings)
   • SONET - can be configured with redundant rings; has built-in redundancy by rerouting frames