Task Scheduling in cloud environment

Classifying the task based based on the requirements and prioritizing task executions on VMs based on the their current workload.

Instructions to run

Lexical Anaysis

• A lexical analyzer analyzer.py is used for classifing the task into n categories and a list of scores is generated for each categories.

• To run the analyzer use python3 analyzer.py. The main function analyze's syntax is as follows.

  analyze(directory=False, directory_name=None, output=False, 
  output_file_name=None, file_name=None)
  

  Arguements:

  directory (boolean) : If set to True it analyzes all the programs in the directory, directory_name argument must be given

  directory_name (string) : If directory is set to True, the name of the directory must be given.

  output (boolean) : Set to True if output is needed, False otherwise.

  output_file_name (string) : File name of output if output flag is set to True.

  file_name (string) : If directory is set to False, then a file name must be given.

• The programs outputs two lists of scores. One is a task specific score for different resources such as (memory, compute, network, etc..), and the other is a resource specific score comparing all programs (if multiple exits).

VM utilisation

The utilsation of each resources in a VM running a task is obtained and is used in the schediling of task. A live state of resources are used for scheduling.

• A bash script utilization.sh is used to get the current resource usage of the VM running the program. A vm_id is assigned to each VM and in this case is the Machine's username obtained using whoami command.
  • top command is used to get the CPU and memory usage.
  • bwm-ng command is used to get the network usage.
• The output of the previous script is parsed with util_extractor.py.
  read_vm_characteristics is exposed by this file and is used to parse the output from the previous script, given the output file name as input.

Task Allocation

This task allocationn requires a live table of resource usage(or available free resource) of each VMs to which tasks needed to be scheduled. It also requires each task to be analyzed using the lexical analyzer as mentioned before

vm_allocator.py is used for task allocation to VMs. This program exposes the following main functions

• insert_new_vm : Given a utilization file as the one generated by utilization.sh it extracts the usage characteristics and inserts it into the vm_table using update_sorted_lists
• update_sorted_lists : This is used internally by the above function to sort the VMs based on each characteristic and is inserted.
• update_vm : This is used to update the usage characteristic of an existing VM in the VM table.
• independent_Task2Vm : This function schedules a single task and needs two inputs one is a list of task_scores for the task, and the task_name to be put in queue.
• dependent_Task2Vm : This functions allocates VMs for multiple tasks. It requires two inputs. task_scores list for all the tasks and a sorted_task_names which is 2d list that is generated by the function given below.
• create_sorted_task_names : This function given a 2d list of task scores for multiple tasks, genrates a sorted list of task names to be used in dependent_Task2Vm.

Other files

• Sample Programs have been given under /programs directory.
• Three files util0.txt, util1.txt, util2.txt have been generated.

Running the program

Dependencies

• python3

• pip3

  In case of ubuntu

  sudo apt-get install python3-pip
  

• tokenizer module

  pip3 install tokenizer
  

Running

Analyzer

python3 analyzer.py

Util File Parser

python3 util_extractor.py

Allocator

python3 vm_allocator.py

Team

Shreyas Seshadri - 16IT135 
Akhil N kashyap  - 16IT104 
GS Dhanush       - 16IT219