Availability in system design refers to the ability of a system to remain operational and accessible when required. It's a key metric that indicates the reliability and resilience of a system.

Imagine a city's public transportation system. In a well-planned city, if one bus route is temporarily unavailable due to a breakdown or road closure, alternative routes or additional buses are quickly deployed to ensure that commuters can still reach their destinations. This redundancy and quick rerouting keep the system running smoothly despite individual issues.

Similarly, in system design, high availability is achieved by having multiple components—such as redundant servers, network paths, or data centers—so that if one component fails, the others can take over seamlessly. This ensures that users continue to access the service without interruption, just as commuters can still travel even if one bus route is disrupted.

Definition

Availability is the proportion of time a system is functional and accessible. It’s typically measured as a percentage (e.g., 99.9% uptime), reflecting the expected operational time versus downtime.

Importance

High availability ensures that users can access services without interruptions, which is critical for user satisfaction, business continuity, and meeting service level agreements (SLAs).

How is Availability Measured ?

Availability is typically measured as the ration of time a system operational (uptime) to the total time it is expected to be available, usually expressed as a percentage.

Uptime Percentage

• Formula:
  Availability = (Uptime / (Uptime + Downtime)) × 100
  For example, if a system is operational for 99.9% of the time, it means that over a given period, it might experience only a small fraction of downtime.

"Nines" of Availability

• Common Benchmarks:
  • 99% Availability (Two Nines): 3.65 days of downtime per year.
  • 99.9% Availability (Three Nines): About 8.76 hours of downtime per year.
  • 99.99% Availability (Four Nines): Roughly 52.6 minutes of downtime per year.
  • 99.999% Availability (Five Nines): Approximately 5.26 minutes of downtime per year.

The "gold standard" often refers to achieving 99.999% uptime—commonly known as "five nines" availability. This level of availability means that a system is allowed only about 5 minutes of downtime per year, making it highly reliable and suitable for mission-critical applications where even brief outages are unacceptable.

How do we achieve high availability ?

Achieving high availability involves designing your system to remain operational even when individual components fail. Here are several key strategies:

• Redundancy:
  Use multiple instances of critical components (servers, databases, network links) so that if one fails, others can take over without interruption.
• Failover Mechanisms:
  Implement automatic failover processes that detect failures and switch to backup systems quickly. This can include clustering or having hot/standby replicas.
• Load Balancing:
  Distribute incoming requests across multiple servers to prevent any single server from becoming a bottleneck or a single point of failure.
• Data Replication:
  Keep multiple copies of data across different servers or data centers. This ensures that even if one data source fails, the system can continue to operate using replicated data.
• Geographical Distribution:
  Deploy your system across multiple regions or data centers. This helps mitigate the impact of localized failures (e.g., power outages, natural disasters).
• Regular Monitoring and Maintenance:
  Use monitoring tools to continuously check system health, and implement automated alerts and remediation processes to address issues before they affect users.
• Resilient Architecture:
  Design your system to gracefully handle failures by isolating faults and ensuring that the failure of one component doesn’t cascade to others.

Difference between high availability and fault tolerance

High availability and fault tolerance are both strategies aimed at keeping a system running, but they approach the goal in different ways:

• High Availability:
  • Goal: Ensure the system is up and accessible most of the time, minimizing downtime.
  • Approach: Uses redundancy, failover mechanisms, and load balancing so that if one component fails, another can take over quickly—usually with a brief interruption.
  • Example: A web application that switches to a standby server within seconds if the primary server goes down.
• Fault Tolerance:
  • Goal: Allow the system to continue operating without any noticeable interruption, even if some components fail.
  • Approach: Builds redundancy at every critical point, often running duplicate components in parallel so that the failure of one has no impact on the system’s overall operation.
  • Example: A database system with multiple active nodes processing transactions simultaneously, so if one node fails, the others seamlessly continue without any downtime.