Introduction to Computer Resilience
In the modern digital world, an unexpected service interruption (downtime) can cause incalculable economic and reputational damage.
Fault Tolerance vs High Availability
Although often confused, these two concepts express radically different engineering approaches to achieving reliability.
Fault Tolerance guarantees absolute continuity without interruptions, using real-time duplicated hardware.
High Availability aims to minimize downtime by accepting brief automated failover transitions.
The Logic of Redundancy
The cardinal rule for eliminating the Single Point of Failure (SPOF) consists of duplicating every single critical element of the infrastructure.
At the hardware level, redundant power supplies connected to separate power lines and uninterruptible power supplies (UPS) are used.
Fault tolerance of hard disks relies on RAID technology to protect data in case of memory failure.
RAID 1 duplicates data in mirroring on two distinct disks, offering simple but expensive protection.
RAID 5 distributes data and parity blocks across a minimum of three disks, ensuring an optimal balance between capacity and security.
RAID 10 combines the advantages of mirroring and striping to achieve stellar performance and maximum robustness.
Load Balancing and Clustering
To scale web services at the software level, Load Balancers are used to distribute user requests across multiple servers.
Server clusters in Active-Active configuration cooperate simultaneously to redistribute the global computational load.
The Active-Passive configuration involves a backup server ready to take over instantly if the primary server stops responding.
Conclusions
Investing in highly available and fault-tolerant systems is the only way to ensure uninterrupted operation at a global level.