2.5 Flashcards
(14 cards)
Redundancy
Duplication of parts of systems so org can continue to operate even with failures on a system
Disk redundancy
Multipath IO creates other paths to ensure one failure doesn’t stop the data flow
Raid, may lose physical drives
there are parts across other drives
Raid 0 striping without parity, no backups
Raid 1 mirroring duplicates for fault tolerance but requires twice the disk space
Raid 5 striping with parity, fault tolerant only requires additional disk for redundancy
Load balancing
Some servers are active some are on standby depending on traffic volume
Nic teaming
Load balancing fail over
Using both active and standby connections to aggregate bandwidth
Multiple network adapters
Nics talk to each other mulicast and fails of Nic doesnt respond
Power redundancy
Uninterrupted Power supply, has batteries that allow For temp outages
1offline/standby ups enable after power is loss
2. Line-interactive, based on voltage to balance to become optimal fillin in difference
3. Online/double conversion ups, always providing batter power no need to switch on or off
Or use generators using fuel
Dual power supplies
Multiple Power supplies For multiple power sources where each can hadle
100% but distributes 50/50 if needed
Power distribution unit
Multiple Power outlets a rack that connects to ethernet and has monitoring
Replication
Frontend and dat seperate so many frontends can access one database, helps when one frontend fails
San storage area networks specialize in high performance network storage devices
Backup types
File backups (archive bit) full back up, incremental (only files since last incremental back up), differential (all files changed since last full backup)
Backup media
Magnetic tape (100gb to terabytes)
Disk back ups (faster than magnetic canduplicate and compress)
Copy back up image
Nas vs sas
Network arrached storage, file level access need to change full file to fix
Storage area network, block level access, can change single portion
Resiliency
Tearing down old instances and creating new
Non persistence, cloud app instances are constantly built and torn down, snapshots can be used to preserve the complete state of a device, making it possible to revert to various different snapshots
High avaialbity
Redundancy doesn’t mean always available
High avaialbity is always on
But comes with higher
Costs
Order of restoration
Needs to correct order To work,
Database first