Edge computing aims to bring data processing closer to the edge of the network. By hosting services which usually reside in the Cloud closer to the end user, it aims to reduce the communication latencies. Fair scheduling of requests for offloading those services onto Edge nodes is an important challenge. EFS aims to provide a prototype for managing the Edge resources and fairly distributing them between the requested services whilst maintaining the base service of the Edge node.
This is a developing research project and some features might not be stable yet.
A. Madej, N. Wang, N. Athanasopoulos, R. Ranjan, and B. Varghese, "Priority-based Fair Scheduling in Edge Computing," Proceedings of the IEEE International Conference on Fog and Edge Computing (ICFEC), 2020.
All source code, documentation, and related artifacts associated with the EFS open source project are licensed under the Apache License, Version 2.0.
In order to run EFS please follow the steps below:
-
Make the install script executable
chmod +x install.sh
-
Run the install script
./install.sh
Linux might complain that there are Windows characters present in the script. In this case run:
sed -i -e 's/\r$//' install.sh; ./install.sh
-
Configure the config.ini within /root/EFS as you desire. Default provided config explained below:
[SERVER] host = 0.0.0.0 port = 6000 maxqueue = 100000 basecpu = 100000 basemem = 256 maxcpu = 100000 cpuunit = 50000 memunit = 256 portlower = 10000 portupper = 19999 strategy = 0
- host – The IP address to bind the socket to. Leave as 0.0.0.0 to bind to all edge node addresses
- port – The port number used for EFS communication
- maxqueue – The maximum number of jobs allowed to be in the queue
- basecpu – The amount of CPU required by the base service. One core is equal to the value set in maxcpu
- basemem – The amount of RAM memory required by the base service in megabytes
- maxcpu – The CPU scheduler period, best left default as 100 microseconds (100000)
- cpuunit – The value of a single CPU unit which will be assigned to each Docker containers. In this the value corresponds to half a core since 50000/100000 = 0.5
- memunit– The value of a single RAM memory unit which will be assigned to each Docker container in megabytes
- portlower – Denotes the start of the range of ports which can be used for the containers
- portupper – Denotes the last value of the range of ports which can be used for the containers
- strategy – Indicates the scheduling strategy to use. The values to use are: 0 for First Come First Served, 1 for Client Fair, 2 for Priority Fair and 3 for Hybrid
-
Generate the server certificate
openssl req -new -newkey rsa:2048 -days 365 -nodes -x509 -keyout server.key -out server.crt
You will be asked to fill in a series of fields which are mainly self explanatory. However it is important that you specify a Common Name you will remember as this is required by the clients
-
Move the generated key and certificate to the certificates' directory
mv servey.key server.crt /root/EFS/certs/
-
Now you need to generate the client certificates. Replace with the client name
openssl req -new -newkey rsa:2048 -days 365 -nodes -x509 -keyout <client>.key -out <client>.crt
Once again you will have to fill in a series of fields. The Common Name here is important as it's used to indentify the client and should match the naming used for the certificate and key
-
Move the generated key and certificate to the certificates' directory
mv <client>.key <client>.crt /root/EFS/certs/
-
Now append the new client key to the client keys file
cat /root/EFS/certs/<client>.key >> /root/EFS/certs/client.crt
-
Distribute the client key and client and server certificate to the client using your preferred method
-
You can now start EFS
cd /root/EFS/; python3.5 EFS.py