Research Article
Cost-Effective Resource Provisioning for Real-Time Workflow in Cloud
| Symbol | Meaning |
| | A workflow represented by DAG | T{t1, t2, …, tn} | A set of n tasks | | The related workloads of n tasks in T | E{eij} | A set of directed edges between two tasks in | D | The deadline of workflow | Parent(ti) | Parent task of ti | Child(ti) | Child task of ti | | Total cost of workflow | | Total workload of workflow | | Total execution time of workflow | | Execution time of task ti on VMj | RT | Resource occupation time | BT | Resource initial booting time | Ci | Cost of task ti | TT | Data transfer time | BL(d) | A set of tasks in layer d | BD(ti) | The longest path of ti to texit | LMET(d) | The minimum execution time of layer d | CSW | Sub-workflow containing all of critical tasks | NSW | Sub-workflow not containing any of critical tasks | | A set of m types of | VMj(MFLOPSj, PRICEj) | A type of | MFLOPSj | Mega-floating point operations per second of | PRICEj | Price of | τ | Billing cycle defined by VMS provider | Pj | Computational performance of | Cj | Cost of resource | X{x1, x2, …, xm} | VMs vector of provisioning results | xj | Number of billing cycles of | fk(W) | First k types of VMs satisfying the workload |
|
|