Advances, Systems and Applications
From: Burstiness-aware service level planning for enterprise application clouds
Symbol | Definition |
---|---|
A | Number of applications |
A M a | Arrival process model for application a |
a max | Application with the top most SLO violation percentage across all A applications |
C a,n,t | Number of resource instances allocated to tier n of application a at interval t |
\(C^{max}_{n,t}\) | Maximum number of resources available for allocation to all applications at tier n in resource allocation interval t |
D a,n | Mean service demand of application a at tier n |
F a | Distribution of number of requests per session for application a |
I a | Index of dispersion of session inter-arrival time for application a |
N a | Number of application tiers for application a |
n bottleneck | Bottleneck tier of an application |
P e n a | SLO violation penalty cost of application a |
R a | Mean request response time of application a predicted by the performance model over the planning horizon |
R a,t | Mean request response time of application a predicted by the performance model over interval t |
R T a | Target mean request response time of application a over the planning horizon |
S a | Number of user sessions in the workload scenario of application a |
S C V a | SCV of session inter-arrival time for application a |
S L O a | SLO of application a |
S M a | Service process model of application a |
T | Number of resource allocation intervals of the planning horizon |
t max | Resource allocation interval with the maximum mean request response time over the planning horizon |
V a | SLO violation percentage of application a |
W | Set of workload scenarios of all applications |
W a | Workload scenario of application a |
Z a | Distribution of think time between session requests in application a |
λ a | Mean session arrival rate for application a |