Redundancy: Having additional systems in case the primary ones fail.
High Availability
Ability of a service to be continuously available by minimizing the downtime to the lower amount possible.
Load Balancing: Process of distributing workloads across multiple computing resources.
Clustering: Use of multiple computers/storage devices/redundant network connections that all work together as a single system to provide high levels of availability, reliability, and scalability. Provides redundancy in event of system failure.
Redundancy: Duplication of critical components or functions of a system with the intention of increasing the reliability of the system. Has to be designed into the architecture.
Data Redundancy
Redundant Array of Independent Disks (RAID): Combines multiple physical storage devices into a recognized single logical storage device.
RAID 0: Provide data striping across multiple disks. For performance without fault tolerance concerns (there is no redundancy in RAID 0).
RAID 1: Mirrors data for redundancy across two drives or SSDs. Has fault tolerance, but only supports using two physical devices. Can lose one storage device without downtime.
RAID 5: Stripes data with parity. Uses at least three storage devices. Parity means in event of failure, remaining disks can reconstruct data on-the-fly. You can hotswap the failed drive without shutting down the server. Can lose one storage device without downtime.
RAID 6: Uses data striping with double parity. At least 4 storage devices. Can lose up to two storage devices without downtime.
RAID 10: Combined RAID 1 and RAID 0; striped array of mirrored arrays. Supports even-numbered amounts of storage devices, starting at minimum of 4. Provides speed and redundancy. Can lose up to two storage devices (as long as they are not in the same array).
Failure-Resistant
Use of redundant storage to withstand hardware malfunctions without data loss (e.g., RAID 1 and RAID 10).
Fault-Tolerant
Allows system to maintain operations in event of downtime (e.g., RAID 1, 5, 6, and 10).
Disaster-Tolerant
Protects data from catastrophic events. Must have two independent zones that always have full access to all data (e.g., RAID 1 and RAID 10).
Capacity Planning
Crucial strategic planning to meet future demands cost-effectively.
Things to consider:
People: Analyze current personnel skills and forecast future needs. Focus on having correct number of skilled individuals to meet organizational goals.
Technology: Assessing current resources, utilization, and anticipating future technological needs.
Infrastructure: Involves considering physical space and utilities to support organizational operations.
Processes: Aims to optimize business processes to handle demand fluctuations (e.g., process automation, outsourcing, etc.).
Powering Data Centers
Power State
Surges: Small and unexpected increase in amount of voltage being provided.
Spikes: Short transient voltage usually caused by a short circuit, tripped circuit breaker, power outage, or lightning strike.
Sags: Small and unexpected decrease in amount of voltage that is being provided.
Undervoltage events: Occurs when voltage is reduced to lower levels and usually occurs for a longer period of time.
Full power loss events: Occurs when there is a total loss of power for a given period of time. Can be followed by a power spike.
Defenses
Line Conditioner: Used to overcome any minor fluctuations in the power being received by the given system. Protects against surges, sags, and some undervoltage events.
Uninterruptible Power Systems (UPS): Provides emergency power to a system when the normal input power source has failed. Also functions as a line conditioner. Most UPS units only provide 15 or 60 minutes, so they are more for just preventing data loss.
Generator: We use this to handle extended full power loss events.
Portable gas-engine: Cheap, loud, hard to maintain, and portable.
Permanently installed: Use natural gas and can power an entire building for hours/days/weeks.
Battery inverter: Quieter and least maintenance, but lower power for less time.
Power Distribution Center (PDC): Central hub where power is received and distributed to all systems in the data center. Does load balancing and monitoring of power.
Data Backups
Never keep all data on one single device or server.
Data Backup: Process of creating duplicate copies of digital information to protect against data loss, corruption, or unavailability.
Onsite v. Offsite
Onsite: Data located in your own office/datacenter. Vulnerable to environmental disaster.
Offsite: Data stored at geographically separate locations from the primary data source to provide protection against physical disasters and to ensure data continuity.
Frequency
How much data is the company willing to lose? Consider the RPO of the organization to confirm the backup plan preserves the necessary data.
Encryption on Backups
Fundamental safeguard that protects the backup data from unauthorized access and potential breachers. Use data-at-rest and data-in-transit encryption.
Snapshots
Point-in-time copies of data that capture a consistent state that is essentially a frozen in time copy. Only stores changes, so very fast and efficient.
Recovery
Used to regain access to data in event of data loss or system failure.
Steps: Selection of the backup, Initiating the recovery process, Data validation, Testing and validation, Documentation and reporting, Notification.
Replication
Making real-time or near-real-time copies of the data. Good for high availability.
Journaling
Maintaining a meticulous record of every change made to an organizations data over time. Good for databases and critical systems.
Continuity of Operations Plan
Ensures an organizations ability to recover from disruptive events or disasters.
Business Continuity Plan (BCP)
Addresses response to disruptive events. Preventative actions and recovery steps for various threats, not limited to technical disruptions (Incident).
Disaster Recovery Plan (DRP)
Subset of BC Plan, focusing on how to resume operations swiftly after a disaster. DRP may include cloud-based solutions (Disaster).
Management
Senior management takes charge of BCP/DRP development. A Business Continuity Coordinator is appointed to lead a Business Continuity Committee. The BC Committee should include representatives from different departments.
Redundant Site Considerations
Alternative site for backup in case primary location encounters a failure or interruption.
Hot Site: Fully equipped backup facility ready to swiftly take over in case of a primary site failure or disruption.
Warm Site: Partially equipped backup that can become operational within days of a primary site disruption.
Cold Site: Site with no immediate equipment or infrastructure. Can get prepared in one to two months.
Mobile Site: Versatile site that utilizes independent and portable units like trailers or tents to deliver recovery capabilities. Key benefit is self sufficiency.
Virtual Site: Utilizes cloud-based environments and offers highly flexible approach.
Platform Diversity
Vital aspect in redundant site design that uses different platforms to prevent single points of failure in disaster recovery. Cloud-Provider Platform Diversity entails spreading resources across multiple cloud providers or regions, reducing the risk of single platform outage.
Resilience and Recovery Testing
Resilience Testing: Assesses the system’s capacity to endure and adjust to disruptive occurrences.
Recovery Testing: Evaluates system’s ability to return to regular functioning following a disruptive event.
Methods:
Tabletop Exercise: Simulate discussion to improve crisis readiness without deploying resources.
Failover Test: Verifies seamless system transition to a backup for uninterrupted functionality during disasters.
Simulation: Computer-generated representations of real-world scenarios. Allows responders to perform actions in a virtualised environment.
Parallel Processing: Replicates data and processes onto a secondary system, which is run in parallel.