High Availability
Summary
PDF p.194High availability (HA) ensures IT systems remain operational with minimal downtime. It involves designing hardware, servers, networking, datacenters, and physical locations for fault tolerance and redundancy. HA setups use redundant components and failover mechanisms to maintain continuous operation.
In plain words
Supplementary — not from your PDFHigh availability (HA) keeps systems running with minimal downtime by using redundant parts, clusters, network redundancy, resilient datacenters and geographic diversity. Availability is expressed in 'nines' (99.999% allows about 5 minutes of downtime a year). Scalability means growing capacity (scale out = more nodes, scale up = bigger nodes); elasticity means adjusting in real time. Recovery sites can be hot (ready immediately), warm (needs latest data loaded) or cold (an empty building).
Detailed explanation
PDF p.194-
High Availability Concepts
- Definition: Ensures systems remain operational and accessible with minimal downtime.
- Redundant Components: Power supplies, hard drives, network interfaces.
- Server Clusters: Automatic failover from primary to secondary servers.
- Networking Redundancy: Switches, routers, load balancers.
- Datacenters: Redundant power sources, cooling systems, backup generators.
- Geographic Diversity: Deploy datacenters in diverse locations to mitigate large-scale events.
-
Measuring Availability
- Uptime and Downtime: Measured over a defined period (e.g., one year).
- Maximum Tolerable Downtime (MTD): Expresses availability requirement.
- "Nines" Term: Describes availability (e.g., 99.9999% for six-nines).
- Downtime Calculation: Sum of scheduled service intervals plus unplanned outages.
-
Scalability and Elasticity
-
Scalability: Increase resources to meet demand within similar cost ratios.
- Scale Out: Add more resources in parallel.
- Scale Up: Increase power of existing resources.
- Elasticity: Handle changes in demand in real time without performance loss.
-
Scalability: Increase resources to meet demand within similar cost ratios.
-
Fault Tolerance and Redundancy
- Fault Tolerance: System continues to provide service despite failures.
- Redundant Components: Allow recovery from component failures.
-
Site Considerations
- Alternate Processing/Recovery Sites: Provide similar level of service.
- Failover: Redundant component or site takes over functionality.
-
Site Resiliency: Hot, warm, or cold sites.
- Hot Site: Immediate failover, operational equipment with live data.
- Warm Site: Requires loading latest data set.
- Cold Site: Empty building, longer setup time.
-
Geographic Dispersion
- Definition: Distribution of recovery sites across different locations.
- Purpose: Minimize impact of regional disasters.
-
Cloud as Disaster Recovery (DR)
- Cost Efficiency: Affordable redundancy and backup options.
- Scalability: Incorporate redundant capabilities without over-provisioning.
- Geographic Diversity: Protect against regional outages.
- Faster Deployment: Quick setup of redundant systems.
- Simplified Management: Tools and services reduce complexity.
- Improved Security and Compliance: Meet regulatory requirements.
-
Testing Redundancy and High Availability
- Load Testing: Validate performance under expected or peak loads.
- Failover Testing: Ensure seamless transition between primary and secondary infrastructure.
- Monitoring Systems: Detect and respond to failures and performance issues.
Important terms
taken from the text above- High Availability Concepts
- Ensures systems remain operational and accessible with minimal downtime.
- Redundant Components
- Power supplies, hard drives, network interfaces.
- Server Clusters
- Automatic failover from primary to secondary servers.
- Networking Redundancy
- Switches, routers, load balancers.
- Datacenters
- Redundant power sources, cooling systems, backup generators.
- Geographic Diversity
- Deploy datacenters in diverse locations to mitigate large-scale events.
- Uptime and Downtime
- Measured over a defined period (e.g., one year).
- Maximum Tolerable Downtime (MTD)
- Expresses availability requirement.
- "Nines" Term
- Describes availability (e.g., 99.9999% for six-nines).
- Downtime Calculation
- Sum of scheduled service intervals plus unplanned outages.
- Scale Out
- Add more resources in parallel.
- Scale Up
- Increase power of existing resources.
- Elasticity
- Handle changes in demand in real time without performance loss.
- Fault Tolerance
- System continues to provide service despite failures.
- Alternate Processing/Recovery Sites
- Provide similar level of service.
- Failover
- Redundant component or site takes over functionality.
- Site Resiliency
- Hot, warm, or cold sites.
- Hot Site
- Immediate failover, operational equipment with live data.
- Warm Site
- Requires loading latest data set.
- Cold Site
- Empty building, longer setup time.
- Geographic Dispersion
- Distribution of recovery sites across different locations.
- Cost Efficiency
- Affordable redundancy and backup options.
- Faster Deployment
- Quick setup of redundant systems.
- Simplified Management
- Tools and services reduce complexity.
- Improved Security and Compliance
- Meet regulatory requirements.
- Load Testing
- Validate performance under expected or peak loads.
- Failover Testing
- Ensure seamless transition between primary and secondary infrastructure.
- Monitoring Systems
- Detect and respond to failures and performance issues.
Examples & real-world scenarios
Supplementary — not from your PDF- A hot site with live data replicated from the primary.
- Dual network links from different ISPs.
- Cloud DR replicating servers into another region.
Scenario
A business can tolerate at most 4 hours of downtime but only has a cold site, which takes weeks to equip. It needs a warm or hot site (or cloud DR) to meet its MTD.
Common mistakes
Supplementary — not from your PDF- Mixing up hot, warm and cold sites.
- Mixing up scale up (more power per node) and scale out (more nodes).
Practical skills
Supplementary — not from your PDF- Convert an availability percentage into allowed downtime.
- Choose a recovery site type from an MTD.
What I should remember
Key Points PDF p.194-
High Availability Concepts
- Redundant Components: Ensure continuous operation.
- Server Clusters: Automatic failover.
- Networking Redundancy: Maintain connectivity.
- Datacenters: Redundant power and cooling.
-
Measuring Availability
- Uptime/Downtime: Defined period.
- MTD: Availability requirement.
- "Nines" Term: Availability description.
-
Scalability and Elasticity
- Scalability: Increase resources.
- Elasticity: Handle demand changes.
-
Fault Tolerance and Redundancy
- Fault Tolerance: Continue service despite failures.
- Redundant Components: Recovery from failures.
-
Site Considerations
- Alternate Sites: Processing/recovery.
- Failover: Redundant component/site.
- Site Resiliency: Hot, warm, cold.
-
Geographic Dispersion
- Recovery Sites: Different locations.
- Purpose: Minimize disaster impact.
-
Cloud as DR
- Cost Efficiency: Affordable redundancy.
- Scalability: Redundant capabilities.
- Geographic Diversity: Protect against outages.
- Faster Deployment: Quick setup.
- Simplified Management: Reduce complexity.
- Security and Compliance: Meet requirements.
-
Testing Redundancy and HA
- Load Testing: Validate performance.
- Failover Testing: Ensure seamless transition.
- Monitoring Systems: Detect/respond to issues.