From: A scheduling algorithm to maximize storm throughput in heterogeneous cluster
Symbols | Descriptions |
---|---|
\(N_{T_i}\) | Number of available machines with type \(T_i\) |
\(N_{C_i}\) | Total number of instances of component i |
m | Total number of worker nodes (machines) |
n | Number of topology’s components |
\(PT_{iw}\) | Processing throughput of task i running on machine w |
\(MAC_w\) | Available CPU capacity of machine w |
\(TCU_{ij}\) | Occupied CPU utilization by ith task on machine j |
\(e_{ij}\) | Average tuple execution time of task i on machine j |
\(MET_{ij}\) | Miscellaneous execution time of Storm for task i running on machine j |
\(R_0\) | Topology initial input rate |
\(\alpha _i\) | Tuple division ratio of ith component |
\(IR_i\) | Input rate of ith task |
\(PR_i\) | Processing rate of ith task |
\(OR_i\) | Output rate of ith task |