The ORIAN Heterogeneous Resource Management Platform

FPT '19 Paper

Demo Videos:

1. Decision Making

2. Profiling

Instructions for use:

The instructions below are how to run the ORIAN platform on Imperial's heterogeneous architecture with access to Maxeler DFE nodes and multi-CPU nodes. To run on another platform, add modified config files to platform/resource_managers/swagger_server/config_files with IP addresses of local machines, and change the paths in the below commands as necessary.

---

To run prototype platform 1 (one heterogeneous node with CPU and DFE children):

---

1. Run the database

cd /platform/Docker

./run-oriain.sh -p 8080

cd platform/database/

rm impl.db task.db // clear impls and tasks

python3 db_prototype.y

On another tab:

cd /platform/Docker

./run-orian.sh

cd platform/database/

python3 setup_dbs.py platform1

2. Run the heterogeneous node manager

cd /platform/Docker

./run-oriain.sh -p 8081

cd platform/resource_managers

python3 -m swagger_server --rm_type HNodeResourceManager --cfg swagger_server/config_files/hnode_cfg.json -p 8081

3. Run the CPU manager

cd /platform/Docker

./run-oriain.sh -p 8082

cd platform/resource_managers

python3 -m swagger_server --rm_type CPUResourceManager --cfg swagger_server/config_files/cpu_cfg.json -p 8082

4. Run the DFE manager

cd /platform/Docker

./run-oriain.sh -p 8083

cd platform/resource_managers

python3 -m swagger_server --rm_type DFEResourceManager --cfg swagger_server/config_files/dfe_cfg.json -p 8083

---

To run prototype platform 2 (cluster with 2 heterogeneous nodes):

---

1. Run the database

cd /platform/Docker

./run-oriain.sh -p 8080

cd platform/database/

rm impl.db task.db // clear impls and tasks

python3 db_prototype.y

On . another tab:

cd /platform/Docker

./run-orian.sh

cd platform/database/

python3 setup_dbs.py platform2

2. Run the cluster manager:

cd /platform/Docker

./run-oriain.sh -p 8087

cd platform/resource_managers

python3 -m swagger_server --rm_type ClusterResourceManager --cfg swagger_server/config_files/cluster_cfg.json -p 8087

3. Run both heterogeneous node managers

cd /platform/Docker

./run-oriain.sh -p 8081

cd platform/resource_managers

python3 -m swagger_server --rm_type HNodeResourceManager --cfg swagger_server/config_files/maxnode_hnode_cfg.json -p 8081

On another tab:

cd /platform/Docker

./run-oriain.sh -p 8084

cd platform/resource_managers

python3 -m swagger_server --rm_type HNodeResourceManager --cfg swagger_server/config_files/maia_hnode_cfg.json -p 8084

4. Run Node 1 CPU manager

cd /platform/Docker

./run-oriain.sh -p 8082

cd platform/resource_managers

python3 -m swagger_server --rm_type CPUResourceManager --cfg swagger_server/config_files/maxnode_cpu_cfg.json -p 8082

5. Run Node 1 DFE manager

cd /platform/Docker

./run-oriain.sh -p 8083

cd platform/resource_managers

python3 -m swagger_server --rm_type DFEResourceManager --cfg swagger_server/config_files/maxnode_dfe_cfg.json -p 8083

6. Run Node 2 CPU manager

cd /platform/Docker

./run-oriain.sh -p 8085

cd platform/resource_managers

python3 -m swagger_server --rm_type CPUResourceManager --cfg swagger_server/config_files/maia_cpu_cfg.json -p 8085

7. Run Node 2 DFE manager

cd /platform/Docker

./run-oriain.sh -p 8086

cd platform/resource_managers

python3 -m swagger_server --rm_type DFEResourceManager --cfg swagger_server/config_files/maia_dfe_cfg.json -p 8086