System Resilience, High Availability, Quality of Service, and Fault Tolerance

Organizations with extreme sensitivity to downtime - medical providers, military and/or intelligence agencies, high-volume online retailers, utilities - have a greater need to ensure BCDR capabilities are comprehensive and effective.

Let's review some techniques for facilitating this practice:

Sufficient spare components

An organization seeking high availability needs to have sufficient components in inventory to replace/repair any affected elements of the environment (or at least those components supporting the critical path).

This is a security concern addressed by logistics and budgeting; having too many spares on hand is an expensive proposition and can negatively impact the organization financially just as much as an outage might.

Clustering

Systems can be combined to provide constant full capacity to the organization when one system/element goes down; this is referred to as "clustering" (storage, processing, and network systems can all be clustered).

This can be viewed as duplication/ replication of the systems in the cluster, and the cluster can enhance normal production through load balancing or merely serve as additional capacity for contingency operations or when widespread temporary scaling is necessary. Common modes of clustering include "active-active"

(where all systems in the cluster operate in normal production, each handling a portion of the operational load and "active-passive" (where at least one of the systems in the cluster doesn't function during normal operations and is only brought online when the normal productions system (s) go into a fail state).

Power

In addition to IT capacity within the environment, the organization must also consider the power requirements for that environment; interruption of sufficient power can lead to loss of availability just as much as a direct impact to the organization's IT/data.

Practical power availability considerations include the following:

• Uninterruptible power supplies (UPS): These devices provide temporary, immediate power during times when utility service is interrupted. Typically the run time for these units is extremely limited relative to the power demand they support. Traditionally based on deep-cycle batteries, other types rely on kinetic energy (Flywheel Energy Storage (FES)) or supercapacitors to store energy, converting it to line power on demand. UPS are only meant to allow users/administrators to conduct formal shutdown procedures on active systems or restoration of line or generator power.

• Generators: A generator is a local power production unit not reliant on utility service. Generators typically run on combustible fuels (gasoline/diesel, natural gas, or propane), and therefore, pose another risk that must be considered in security/BCDR planning (for both fire/ explosions and toxic exhaust). Generators should also be paired with a transfer switch, which is a mechanism that detects loss of utility power and engages the generator automatically to compensate.