Cloud Computing: SLA & Service Quality Metrics

Vocabulary practice flashcards covering Service Level Agreements (SLAs), Service Quality Metrics (SQMs), contractual remedies, availability formulas, and resiliency definitions from Cloud Computing Lecture 03.

Service Level Agreement (SLA)

A formal document that codifies the specific parameters, Quality of Service (QoS) features, minimum performance levels, guarantees, limitations, and remedies required for elements of a cloud service.

SLA Corrections

Specific remedies stated in an SLA that codify the mandatory actions a service provider must take to prevent a recurrence of an SLA failure.

SLA Penalties

Remedies defined in an SLA for non-compliance when agreed thresholds are not met, which often take the form of financial service credits.

Availability Rate Metric

A service quality metric expressing the percentage of service up-time, calculated as: $\frac{\text{total up-time}}{\text{total time}}$.

Outage Duration Metric

An availability metric that measures the continuous duration of a single outage per event, calculated as: $\text{date/time of outage end} - \text{date/time of outage start}$.

Mean-Time Between Failures (MTBF)

A reliability metric representing the expected time between consecutive service failures, calculated as: $\frac{\text{total normal operational period duration}}{\text{number of failures}}$.

Reliability Rate Metric

A service quality metric expressing the percentage of successful service outcomes under pre-defined conditions, measured as: $\frac{\text{total number of successful responses}}{\text{total number of requests}}$.

Resiliency

A service quality metric characterizing a system's capacity to maintain operation or recover through redundant implementation, resource replication across physical locations, and disaster recovery systems across design, operational, and recovery phases.

Mean-Time To Switchover (MTSO)

A resiliency metric defined as the expected duration for a resilient system to accomplish full recovery from a server failure, calculated as: $\frac{\text{date/time of recovery} - \text{date/time of failure}}{\text{total number of failures}}$.

Mean-Time System Recovery (MTSR)

A resiliency metric describing the time expected to execute a failover/switchover from a failed service to an active replica in a distinct geographical region, measured as: $\frac{\text{date/time of switchover completion} - \text{date/time of failure}}{\text{total number of failures}}$.

Downtime (Google Apps SLA)

A condition occurring for a domain when there is a server-side error rate greater than $5\%$ during a given period.

Downtime Period (Google Apps SLA)

A timeframe consisting of ten consecutive minutes of Downtime for a domain; intermittent downtime lasting less than ten minutes is not counted.

Monthly Uptime Percentage (Google Apps SLA)

The proportion of operational service time in a month, defined as the total number of minutes in a calendar month minus minutes of Downtime from all Downtime Periods, divided by the total number of minutes in that calendar month.

Scheduled Downtime (Google Apps SLA)

Planned outages where Google gives the customer at least $5$ days advance notice; capped at a maximum of $12$ hours per calendar year and excluded from Downtime calculations.

Perimeter Security in Cloud Computing

A traditional infrastructure-protection model that does not exist in the cloud; security efforts must secure the data traffic rather than the underlying physical infrastructure.

Variable $n$ (Availability Formula)

The variable in the availability calculation representing the total number of servers, determined by the equation: $n = 1 + \text{number of duplicates}$.

Variable $c$ (Availability Formula)

The variable in the availability calculation representing the probability or percentage likelihood of encountering a server loss within a given measurement period.

Variable $d$ (Availability Formula)

The variable in the availability calculation representing the expected downtime (in hours) caused by the failure or loss of a server.

Variable $p$ (Availability Formula)

The variable representing the total duration of the measurement period in hours, typically set to $8760$ hours for annual calculations.

Full Availability Formula (with External Outages)

The mathematical formula for expected availability ($a$) incorporating server failure probabilities, duplicate redundancy, and other external outage hours ($o$): $a = \frac{p - \left(\frac{c \times d^n}{p^{n-1}} + o\right)}{p}$.