Availability

Availability is the percentage of time a system is operational and accessible.

The Formula

Availability = (Uptime / Total Time) × 100%

Example:
Uptime = 364 days 23 hours 56 minutes = 364.9972 days
Total Time = 365 days
Availability = (364.9972 / 365) × 100% = 99.999%

The Nines of Availability

Nines	Downtime/Year	Downtime/Month	Use Case
99%	3.65 days	7.31 hours	Non-critical
99.9%	8.76 hours	43.8 min	Internal tools
99.99%	52.6 min	4.38 min	Web applications
99.999%	5.26 min	26.3 sec	Financial systems
99.9999%	31.5 sec	2.63 sec	Critical infrastructure

Downtime Cost Example

System: E-commerce platform
Revenue: $100,000/hour

99.9% availability → 8.76 hours downtime → $876,000 lost
99.99% availability → 52.6 min downtime → $87,700 lost
99.999% availability → 5.26 min downtime → $8,770 lost

Why Availability Matters

User Expectations:
- 99.9% = Users notice occasional downtime
- 99.99% = Users rarely notice downtime
- 99.999% = Users never notice downtime

Business Impact:
- Downtime = Lost revenue
- Downtime = Lost trust
- Downtime = SLA penalties

Common Availability Numbers

Service	Target Availability
Google Search	99.99%
AWS EC2	99.99%
Azure	99.95%
Netflix	99.99%
Banking	99.999%

Redundancy eliminates single points of failure by having backup components.

Types of Redundancy

Redundancy
├── Hardware Redundancy
│   ├── Duplicate servers
│   ├── RAID storage
│   └── Redundant power supplies
├── Software Redundancy
│   ├── Multiple instances
│   ├── Replicated data
│   └── Backup services
├── Network Redundancy
│   ├── Multiple network paths
│   ├── Redundant switches
│   └── Multiple ISPs
└── Geographic Redundancy
    ├── Multiple data centers
    ├── Multi-region deployment
    └── CDN distribution

Server Redundancy

Without Redundancy:        With Redundancy:
┌──────────┐              ┌──────────┐ ┌──────────┐
│ Server 1 │              │ Server 1 │ │ Server 2 │
│ (SPOF)   │              │ (Active) │ │ (Standby)│
└──────────┘              └──────────┘ └──────────┘
      │                        │            │
      │                        └──────┬─────┘
      ▼                               ▼
  Failure = Downtime            Failure = No downtime

Data Redundancy

Primary Database ──── Replication ────→ Replica 1
      │                                 Replica 2
      └──── Replication ────→ Replica 3

If Primary fails:
- Promote Replica to Primary
- Continue operations with minimal downtime

Geographic Redundancy

                    ┌─────────────────────┐
                    │    Load Balancer     │
                    └──────────┬──────────┘
               ┌───────────────┼───────────────
               │               │               │
        ┌──────▼──────┐ ┌─────▼───────┐ ┌────▼──────┐
        │  US-East    │ │  US-West    │ │  EU-West  │
        │  Data Center│ │  Data Center│ │  Data Ctr │
        └─────────────┘ └─────────────┘ └───────────┘

Region Failure → Traffic routed to other regions

Failover is the automatic switching to a backup system when the primary fails.

Types of Failover

Type	Description	Recovery Time
Automatic	System detects failure and switches	Seconds to minutes
Manual	Operator triggers failover	Minutes to hours
Semi-automatic	Auto-detect, manual trigger	Minutes

Active-Passive Failover

Normal Operation:
┌─────────────┐     ┌─────────────┐
│   Active    │     │   Passive   │
│  (serving)  │     │  (waiting)  │
└──────┬──────┘     └──────┬──────┘
       │                   │
       └────────┬──────────┘
                │
         ┌──────▼──────┐
         │   Traffic   │
         └─────────────┘

After Failure:
┌─────────────┐     ┌─────────────┐
│   Active    │     │   Passive   │
│   (DOWN)    │     │  (NOW ACTIVE)│
└──────┬──────┘     └──────┬──────┘
       │                   │
       └────────┬──────────┘
                │
         ┌──────▼──────┐
         │   Traffic   │
         └─────────────┘

Active-Active Failover

Both servers actively serve traffic:

┌─────────────┐     ┌─────────────┐
│  Active 1   │     │  Active 2   │
│  (serving)  │     │  (serving)  │
└──────┬──────┘     └──────┬──────┘
       │                   │
       └────────┬──────────┘
                │
         ┌──────▼──────┐
         │   Traffic   │
         └─────────────┘

If one fails, other continues serving 100% of traffic

Health Check Configuration

HealthCheck:
  Protocol: HTTP
  Path: /health
  Port: 8080
  Interval: 10 seconds
  Timeout: 5 seconds
  UnhealthyThreshold: 3  # Failures before marking unhealthy
  HealthyThreshold: 2    # Successes before marking healthy

Failover Best Practices

1. Test failover regularly: Don't wait for real failures
2. Monitor failover events: Alert on unexpected failovers
3. Use circuit breakers: Prevent cascading failures
4. Implement graceful degradation: Reduce functionality under load
5. Document runbooks: Know how to handle each failure scenario