The implementation of TEEMS, a Trusted Execution Environment based Metadata-protected Messaging System
Ian Goldberg
7b90bbebdb
Report results meaningfully when SHRINK_TO_MEM and/or OVERLOAD_CORES are set to 1
|
10 mesi fa | |
|---|---|---|
| App | 11 mesi fa | |
| Client | 11 mesi fa | |
| Enclave | 11 mesi fa | |
| Untrusted | 11 mesi fa | |
| docker | 10 mesi fa | |
| Makefile | 11 mesi fa | |
| README | 1 anno fa | |
| core-time-tradeoff | 1 anno fa | |
| core_allocator.py | 11 mesi fa | |
| epc_probe.py | 11 mesi fa | |
| gen_enclave_config.py | 11 mesi fa | |
| gen_manifest.py | 11 mesi fa | |
| logs_to_csv.py | 1 anno fa | |
| mkzip | 1 anno fa | |
| plot_traces | 1 anno fa | |
| run_all_experiments.py | 11 mesi fa | |
| run_experiments.py | 10 mesi fa | |
| short_exp.py | 10 mesi fa |
README
Source code for "TEEMS: Trusted Execution Environment based
Metadata-protected Messaging System"
This code produced the data reported in Section 6 of our paper.
You will need:
- A machine with one or more SGX2-enabled processors
- The SGX SDK (https://github.com/intel/linux-sgx) installed (we used
version 2.22)
- The "numactl" program installed (even if you don't have a NUMA
machine)
The code is designed to run on a single machine with two 40-core
processors (Intel 8380). We allocate 36 cores on each processor for the
servers, and the last 4 cores on each processor for the clients.
If you have a different configuration, change the cores assigned to
clients by setting "prefix" in Client/clientlaunch and the -t option to
clientlaunch in run_experiments.py. Change the cores assigned to
servers in the gen_manifest.py program. Change the maximum number of
servers for a given number of cores in the M_MAX dictionary in
run_experiments.py.
Decide what experiments you want to run:
- Set the numbers of users (N), the numbers of servers (M), and the
numbers of cores per server (T) in run_experiments.py.
- Set whether you want to test the private or public channel by setting
"private_routing:" in gen_manifest.py to True or False respectively.
Build and run the code, and collect the output:
- make
- ./run_experiments.py
- ./logs_to_csv.py Experiments_test results.csv
Metadata-protected Messaging System"
This code produced the data reported in Section 6 of our paper.
You will need:
- A machine with one or more SGX2-enabled processors
- The SGX SDK (https://github.com/intel/linux-sgx) installed (we used
version 2.22)
- The "numactl" program installed (even if you don't have a NUMA
machine)
The code is designed to run on a single machine with two 40-core
processors (Intel 8380). We allocate 36 cores on each processor for the
servers, and the last 4 cores on each processor for the clients.
If you have a different configuration, change the cores assigned to
clients by setting "prefix" in Client/clientlaunch and the -t option to
clientlaunch in run_experiments.py. Change the cores assigned to
servers in the gen_manifest.py program. Change the maximum number of
servers for a given number of cores in the M_MAX dictionary in
run_experiments.py.
Decide what experiments you want to run:
- Set the numbers of users (N), the numbers of servers (M), and the
numbers of cores per server (T) in run_experiments.py.
- Set whether you want to test the private or public channel by setting
"private_routing:" in gen_manifest.py to True or False respectively.
Build and run the code, and collect the output:
- make
- ./run_experiments.py
- ./logs_to_csv.py Experiments_test results.csv