| Name | SLAMBUC JSON |
| Version |
0.2.1
JSON |
| download |
| home_page | |
| Summary | Serverless Layout Adaptation with Memory-Bounds and User Constraints |
| upload_time | 2023-09-06 16:41:10 |
| maintainer | |
| docs_url | None |
| author | Janos Czentye |
| requires_python | >=3.10 |
| license | Apache 2.0 |
| keywords |
cloud
serverless
ilp
dp
tree
|
| VCS |
 |
| bugtrack_url |
|
| requirements |
No requirements were recorded.
|
| Travis-CI |
No Travis.
|
| coveralls test coverage |
No coveralls.
|
# Serverless Layout Adaptation with Memory-Bounds and User Constraints (SLAMBUC)
[](https://pypi.org/project/SLAMBUC/)
[](https://pypi.org/project/SLAMBUC/#history)
[](https://pepy.tech/project/slambuc)
[](LICENSE)
[](https://github.com/hsnlab/SLAMBUC/actions/workflows/python-alg-tests.yml)
[](https://github.com/hsnlab/SLAMBUC/actions/workflows/python-alg-validation.yml)
Collection of graph partitioning algorithms implemented in Python for composing cloud-native
applications from standalone serverless functions in a cost-efficient and latency-constrained manner.
## Overview
In the context of serverless computing, function fusion is a novel, high-level approach to improve
performance and at the same time reduce the operational cost of serverless applications consisting
of stateless, ephemeral functions. This is achieved by grouping, encompassing, and assembling connected
FaaS functions into separate composite components representing the deployable software artifacts that
are provisioned by the serverless frameworks in the same way as other single functions.
In addition, different user-defined Quality of Service (QoS) constraints should be also taken into
account, e.g., overall response time of the application or an end-to-end latency constraint on the critical
path in the application's call graph.
Under the hood, this problem can be formalized as the partitioning of the application call graph (DAG)
into disjoint, connected subgraphs in a cost-efficient manner, while specific requirements imposed by
the user and the platform (flavors) itself need to be satisfied.
In this package, we designed, implemented, and collected various partitioning algorithms tailored to
tree-shape serverless applications with different runtime complexity, considering communication
parameters and requirements. Our main objective is to find the cost-optimal grouping of functions
concerning node and edge-weighted trees and cost/memory/latency models based on public cloud frameworks,
whereas each flavor imposes an upper limit on the available operative memory.
Moreover, a user-given latency constraint has to be fulfilled on the tree's critical path, which is
defined as the subchain between the first/front-end function and a predefined leaf node.
## Installation
### Environment
Our implementations require Python3.10 or above. The following code snippet can be used to set up the
latest Python environment on Ubuntu.
```bash
sudo add-apt-repository -y 'ppa:deadsnakes/ppa' && sudo apt update
sudo apt install python3.11-dev python3.11-dev
sudo curl -sS https://bootstrap.pypa.io/get-pip.py | sudo python3.11
```
### SLAMBUC package
The easiest way to get our algorithms collected in [SLAMBUC](slambuc) is to install the package from
[PyPI repository](https://pypi.org/project/SLAMBUC/).
```bash
python3.11 -m pip install slambuc
```
However, for the latest changes, it can be installed directly from GitHub as follows.
```bash
python3.11 -m pip install --no-cache-dir git+https://github.com/hsnlab/SLAMBUC.git
```
Tree plotting relies on networkx's internal plotting feature that generates a layout based on the
[graphviz tool and its python frontend](https://pygraphviz.github.io/documentation/stable/install.html).
Thus, in that case, the related dependencies must be installed first.
```bash
sudo apt-get install graphviz graphviz-dev
python3.11 -m pip install pygraphviz
```
External solvers can also be used in LP-based algorithms that require the given solver packages to be
preinstalled and available for the [PuLP frontend](https://github.com/coin-or/pulp). Currently,
the following solvers are tested.
* CBC (default, packaged with PuLP)
* GLPK (see installation [here](https://coin-or.github.io/pulp/main/installing_pulp_at_home.html#linux-installation))
* CPLEX ([installation](https://www.ibm.com/products/ilog-cplex-optimization-studio)
and [setup](https://coin-or.github.io/pulp/guides/how_to_configure_solvers.html#cplex))
It is worth noting that CPLEX's python wrapper [docplex](https://pypi.org/project/docplex/)
(as a replacement for PuLP) is left behind the latest Python version. For using this API, requirements
are prepared separately for **Python3.10**.
```bash
python3.10 -m pip install -U -r requirements_py3.10.txt
```
For solving constrained shortest path problems (CSP), we apply solution methods from
[cspy](https://github.com/torressa/cspy).
### Test harness and performance validation
Our repository contains separate test scripts under the [tests](tests) folder for validating the
input/output formats and call parameters. These codes also serve as examples for using the different
implementations of our package.
For comparative analyses, we also implemented a test harness under [validation](validation)
to automatize test executions with generated test input graphs from [validation/data](validation/data)
and monitor elapsed time and memory demands of tested algorithms initiated as separate sub-processes.
To install additional dependencies, run the following commands.
```bash
python3.11 -m pip install slambuc[tests] # For executing tests
python3.11 -m pip install slambuc[validation] # For using our test harness framework
```
## Usage
Refer to the wiki for [formats, execution parameters, examples, and API documentation](https://github.com/hsnlab/SLAMBUC/wiki).
## Example
```python
from slambuc.alg.tree.ser.pseudo import pseudo_ltree_partitioning
from slambuc.misc.generator import get_random_tree
# Get input parameters
tree = get_random_tree(nodes=10)
params = dict(tree=tree,
root=1,
M=6,
L=450,
cp_end=10,
delay=10)
# Partitioning
res = pseudo_ltree_partitioning(**params)
print(f"Part: {res[0]}, opt. cost: {res[1]}, latency: {res[2]}")
"Part: [[1, 2], [3, 4], [5, 6, 7, 8], [9], [10]], opt. cost: 1252, latency: 449"
```
## Algorithms
Validation results of a subset of our algorithms with a fully-serialized block execution model,
which are executed with our [validation script](tests/validate_algs.py) using different configurations
and a [random-generated input call graph](tests/data/graph_test_tree_ser.gml) of size 10.
Used algorithmic parameters (if applicable):
* Root node ID (root): 1
* Memory limit (M): 6
* Available vCPU count (N): 1
* Critical path's end node ID (cp_end): 10
* Latency limit: (L): **500**
* Platform delay: (delay): 10
* Bidirectional elimination (bidirectional): True
* Cost approximation ratio (Epsilon): 0.0
* Latency violation ratio (Lambda): 0.0
Exact algorithms are configured to yield all optimal solutions (if exists) with the numerating
format `{alg}_{num}`.
Execution results:
| Algorithm | Partitioning | Cost | Latency | Time (s) |
|--------------------|-----------------------------------------------------|--------|------------------------------------|------------|
| GREEDY_0 | [[1, 2, 3], [4, 5, 6, 8, 9, 10], [7]] | 858 | 443 | 0.0235749 |
| GREEDY_1 | [[1, 3, 4, 5], [2], [6, 8, 9, 10], [7]] | 858 | 474 | 0.0235749 |
| GREEDY_2 | [[1, 3, 4, 5], [2], [6, 7, 8, 9], [10]] | 858 | 471 | 0.0235749 |
| ILP_CFG_HYBRID | [[1, 3, 4, 5], [2], [6, 8, 9, 10], [7]] | 858 | 474 | 0.0167496 |
| ILP_MTX | [[1, 3, 4, 5], [2], [6, 8, 9, 10], [7]] | 858 | 474 | 0.0197985 |
| PSEUDO_B | [[1, 2, 3], [4, 5, 6, 8, 9, 10], [7]] | 858 | 443 | 0.00047041 |
| PSEUDO_L | [[1, 3, 4, 5], [2], [6, 7, 8, 9], [10]] | 858 | 471 | 0.00083811 |
| BIFPTAS_L | [[1, 3, 4, 5], [2], [6, 7, 8, 9], [10]] | 858 | 471 | 0.00082326 |
| _BASELINE_NO_PART_ | [[1], [2], [3], [4], [5], [6], [7], [8], [9], [10]] | 1090 | 472 | 9.38e-05 |
| _BASELINE_SINGLE_ | [[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]] | 822 | <span style="color:red">686</span> | 6.718e-05 |
## Development and contribution
If you would like to contribute, add a feature, or just play with the implementations, the development
environment can be set up with the following commands.
```bash
git clone https://github.com/hsnlab/SLAMBUC.git
python3.11 -m pip install -U -r SLAMBUC/requirements.txt
python3.11 -m pip install --no-deps -e SLAMBUC/
# OR
cd SLAMBUC && make install-req && make dev-install
## Remove editing-mode package outside of repo root
python3.11 -m pip uninstall slambuc
# OR
make uninstall
```
## Publications
If you use one of our algorithms published in this package or our test harness, please consider citing
one of our related works.
#### [Polynomial-time algorithms based on chain-based tree partitioning:](https://doi.org/10.1109/noms56928.2023.10154412)
J. Czentye, I. Pelle and B. Sonkoly,
"Cost-optimal Operation of Latency Constrained Serverless Applications: From Theory to Practice,"
_NOMS 2023-2023 IEEE/IFIP Network Operations and Management Symposium_, Miami, FL, USA, 2023, pp. 1-10,
doi: 10.1109/NOMS56928.2023.10154412.
```bibtex
@INPROCEEDINGS{Czentye2022noms,
author = {J{\'{a}}nos Czentye and Istv{\'{a}}n Pelle and Bal{\'{a}}zs Sonkoly},
booktitle = {{NOMS 2023-2023 IEEE/IFIP Network Operations and Management Symposium}},
title = {{Cost-optimal Operation of Latency Constrained Serverless Applications: From Theory to Practice}},
publisher = {{IEEE}},
year = {2023},
month = may,
pages = {1--10},
doi = {10.1109/NOMS56928.2023.10154412}
}
```
#### [Heuristic algorithm for dynamic (re)optimization control loop in edge-could environments:](https://doi.org/10.1109/jiot.2020.3042428)
I. Pelle, J. Czentye, J. Dóka, A. Kern, B. P. Gerő and B. Sonkoly,
"Operating Latency Sensitive Applications on Public Serverless Edge Cloud Platforms,"
in _IEEE Internet of Things Journal_, vol. 8, no. 10, pp. 7954-7972, 15 May, 2021,
doi: 10.1109/JIOT.2020.3042428.
```bibtex
@ARTICLE{Pelle2021jiot,
author = {Pelle, Istv{\'{a}}n and Czentye, J{\'{a}}nos and D{\'{o}}ka, J{\'{a}}nos and Kern, Andr{\'{a}}s and Ger{\H{o}}, Bal{\'{a}}zs P. and Sonkoly, Bal{\'{a}}zs},
journal = {{IEEE Internet of Things Journal}},
title = {{Operating Latency Sensitive Applications on Public Serverless Edge Cloud Platforms}},
publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},
year = {2021},
month = may,
volume = {8},
number = {10},
pages = {7954--7972},
doi = {10.1109/JIOT.2020.3042428}
}
```
#### [Layout optimization for serverless applications over public clouds:](https://doi.org/10.1109/globecom38437.2019.9013988)
J. Czentye, I. Pelle, A. Kern, B. P. Gero, L. Toka and B. Sonkoly,
"Optimizing Latency Sensitive Applications for Amazon's Public Cloud Platform,"
_2019 IEEE Global Communications Conference (GLOBECOM)_, Waikoloa, HI, USA, 2019, pp. 1-7,
doi: 10.1109/GLOBECOM38437.2019.9013988.
```bibtex
@INPROCEEDINGS{Czentye2019globecom,
author = {Czentye, Janos and Pelle, Istvan and Kern, Andras and Gero, Balazs Peter and Toka, Laszlo and Sonkoly, Balazs},
booktitle = {{2019 IEEE Global Communications Conference (GLOBECOM)}},
title = {{Optimizing Latency Sensitive Applications for Amazon's Public Cloud Platform}},
publisher = {{IEEE}},
year = {2019},
month = dec,
pages = {1--7},
doi = {10.1109/GLOBECOM38437.2019.9013988}
}
```
## License
SLAMBUC is an open-source software licensed under [Apache 2.0](LICENSE).
Raw data
{
"_id": null,
"home_page": "",
"name": "SLAMBUC",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.10",
"maintainer_email": "",
"keywords": "cloud,serverless,ilp,dp,tree",
"author": "Janos Czentye",
"author_email": "Janos Czentye <czentye@tmit.bme.hu>",
"download_url": "https://files.pythonhosted.org/packages/6e/ed/3ad8c0488a2f79bec306e32e225b55b563301082fc12c784e56bdc9d5b79/SLAMBUC-0.2.1.tar.gz",
"platform": null,
"description": "# Serverless Layout Adaptation with Memory-Bounds and User Constraints (SLAMBUC)\n\n[](https://pypi.org/project/SLAMBUC/)\n[](https://pypi.org/project/SLAMBUC/#history)\n[](https://pepy.tech/project/slambuc)\n[](LICENSE)\n[](https://github.com/hsnlab/SLAMBUC/actions/workflows/python-alg-tests.yml)\n[](https://github.com/hsnlab/SLAMBUC/actions/workflows/python-alg-validation.yml)\n\nCollection of graph partitioning algorithms implemented in Python for composing cloud-native\napplications from standalone serverless functions in a cost-efficient and latency-constrained manner.\n\n## Overview\n\nIn the context of serverless computing, function fusion is a novel, high-level approach to improve\nperformance and at the same time reduce the operational cost of serverless applications consisting\nof stateless, ephemeral functions. This is achieved by grouping, encompassing, and assembling connected\nFaaS functions into separate composite components representing the deployable software artifacts that\nare provisioned by the serverless frameworks in the same way as other single functions.\nIn addition, different user-defined Quality of Service (QoS) constraints should be also taken into\naccount, e.g., overall response time of the application or an end-to-end latency constraint on the critical \npath in the application's call graph.\n\nUnder the hood, this problem can be formalized as the partitioning of the application call graph (DAG)\ninto disjoint, connected subgraphs in a cost-efficient manner, while specific requirements imposed by\nthe user and the platform (flavors) itself need to be satisfied.\n\nIn this package, we designed, implemented, and collected various partitioning algorithms tailored to\ntree-shape serverless applications with different runtime complexity, considering communication\nparameters and requirements. Our main objective is to find the cost-optimal grouping of functions \nconcerning node and edge-weighted trees and cost/memory/latency models based on public cloud frameworks,\nwhereas each flavor imposes an upper limit on the available operative memory.\nMoreover, a user-given latency constraint has to be fulfilled on the tree's critical path, which is\ndefined as the subchain between the first/front-end function and a predefined leaf node.\n\n## Installation\n\n### Environment\n\nOur implementations require Python3.10 or above. The following code snippet can be used to set up the \nlatest Python environment on Ubuntu.\n\n```bash\nsudo add-apt-repository -y 'ppa:deadsnakes/ppa' && sudo apt update\nsudo apt install python3.11-dev python3.11-dev\nsudo curl -sS https://bootstrap.pypa.io/get-pip.py | sudo python3.11\n```\n\n### SLAMBUC package\n\nThe easiest way to get our algorithms collected in [SLAMBUC](slambuc) is to install the package from\n[PyPI repository](https://pypi.org/project/SLAMBUC/).\n\n```bash\npython3.11 -m pip install slambuc\n```\n\nHowever, for the latest changes, it can be installed directly from GitHub as follows.\n\n```bash\npython3.11 -m pip install --no-cache-dir git+https://github.com/hsnlab/SLAMBUC.git\n```\n\nTree plotting relies on networkx's internal plotting feature that generates a layout based on the\n[graphviz tool and its python frontend](https://pygraphviz.github.io/documentation/stable/install.html).\nThus, in that case, the related dependencies must be installed first.\n\n```bash\nsudo apt-get install graphviz graphviz-dev\npython3.11 -m pip install pygraphviz\n```\n\nExternal solvers can also be used in LP-based algorithms that require the given solver packages to be\npreinstalled and available for the [PuLP frontend](https://github.com/coin-or/pulp). Currently,\nthe following solvers are tested.\n\n* CBC (default, packaged with PuLP)\n* GLPK (see installation [here](https://coin-or.github.io/pulp/main/installing_pulp_at_home.html#linux-installation))\n* CPLEX ([installation](https://www.ibm.com/products/ilog-cplex-optimization-studio)\n and [setup](https://coin-or.github.io/pulp/guides/how_to_configure_solvers.html#cplex))\n\nIt is worth noting that CPLEX's python wrapper [docplex](https://pypi.org/project/docplex/)\n(as a replacement for PuLP) is left behind the latest Python version. For using this API, requirements\nare prepared separately for **Python3.10**.\n\n```bash\npython3.10 -m pip install -U -r requirements_py3.10.txt\n```\n\nFor solving constrained shortest path problems (CSP), we apply solution methods from\n[cspy](https://github.com/torressa/cspy).\n\n### Test harness and performance validation\n\nOur repository contains separate test scripts under the [tests](tests) folder for validating the \ninput/output formats and call parameters. These codes also serve as examples for using the different \nimplementations of our package.\n\nFor comparative analyses, we also implemented a test harness under [validation](validation)\nto automatize test executions with generated test input graphs from [validation/data](validation/data)\nand monitor elapsed time and memory demands of tested algorithms initiated as separate sub-processes.\n\nTo install additional dependencies, run the following commands.\n\n```bash\npython3.11 -m pip install slambuc[tests] # For executing tests\npython3.11 -m pip install slambuc[validation] # For using our test harness framework\n```\n\n## Usage\n\nRefer to the wiki for [formats, execution parameters, examples, and API documentation](https://github.com/hsnlab/SLAMBUC/wiki).\n\n## Example\n\n```python\nfrom slambuc.alg.tree.ser.pseudo import pseudo_ltree_partitioning\nfrom slambuc.misc.generator import get_random_tree\n\n# Get input parameters\ntree = get_random_tree(nodes=10)\nparams = dict(tree=tree,\n root=1,\n M=6,\n L=450,\n cp_end=10,\n delay=10)\n\n# Partitioning\nres = pseudo_ltree_partitioning(**params)\nprint(f\"Part: {res[0]}, opt. cost: {res[1]}, latency: {res[2]}\")\n\"Part: [[1, 2], [3, 4], [5, 6, 7, 8], [9], [10]], opt. cost: 1252, latency: 449\"\n```\n\n## Algorithms\n\nValidation results of a subset of our algorithms with a fully-serialized block execution model,\nwhich are executed with our [validation script](tests/validate_algs.py) using different configurations \nand a [random-generated input call graph](tests/data/graph_test_tree_ser.gml) of size 10.\n\nUsed algorithmic parameters (if applicable):\n * Root node ID (root): 1\n * Memory limit (M): 6\n * Available vCPU count (N): 1\n * Critical path's end node ID (cp_end): 10 \n * Latency limit: (L): **500**\n * Platform delay: (delay): 10\n * Bidirectional elimination (bidirectional): True\n * Cost approximation ratio (Epsilon): 0.0\n * Latency violation ratio (Lambda): 0.0\n\nExact algorithms are configured to yield all optimal solutions (if exists) with the numerating \nformat `{alg}_{num}`.\n\nExecution results:\n\n| Algorithm | Partitioning | Cost | Latency | Time (s) |\n|--------------------|-----------------------------------------------------|--------|------------------------------------|------------|\n| GREEDY_0 | [[1, 2, 3], [4, 5, 6, 8, 9, 10], [7]] | 858 | 443 | 0.0235749 |\n| GREEDY_1 | [[1, 3, 4, 5], [2], [6, 8, 9, 10], [7]] | 858 | 474 | 0.0235749 |\n| GREEDY_2 | [[1, 3, 4, 5], [2], [6, 7, 8, 9], [10]] | 858 | 471 | 0.0235749 |\n| ILP_CFG_HYBRID | [[1, 3, 4, 5], [2], [6, 8, 9, 10], [7]] | 858 | 474 | 0.0167496 |\n| ILP_MTX | [[1, 3, 4, 5], [2], [6, 8, 9, 10], [7]] | 858 | 474 | 0.0197985 |\n| PSEUDO_B | [[1, 2, 3], [4, 5, 6, 8, 9, 10], [7]] | 858 | 443 | 0.00047041 |\n| PSEUDO_L | [[1, 3, 4, 5], [2], [6, 7, 8, 9], [10]] | 858 | 471 | 0.00083811 |\n| BIFPTAS_L | [[1, 3, 4, 5], [2], [6, 7, 8, 9], [10]] | 858 | 471 | 0.00082326 |\n| _BASELINE_NO_PART_ | [[1], [2], [3], [4], [5], [6], [7], [8], [9], [10]] | 1090 | 472 | 9.38e-05 |\n| _BASELINE_SINGLE_ | [[1, 2, 3, 4, 5, 6, 7, 8, 9, 10]] | 822 | <span style=\"color:red\">686</span> | 6.718e-05 |\n\n## Development and contribution\n\nIf you would like to contribute, add a feature, or just play with the implementations, the development\nenvironment can be set up with the following commands.\n\n```bash\ngit clone https://github.com/hsnlab/SLAMBUC.git\npython3.11 -m pip install -U -r SLAMBUC/requirements.txt\npython3.11 -m pip install --no-deps -e SLAMBUC/\n# OR\ncd SLAMBUC && make install-req && make dev-install\n\n## Remove editing-mode package outside of repo root\npython3.11 -m pip uninstall slambuc\n# OR\nmake uninstall\n```\n\n## Publications\n\nIf you use one of our algorithms published in this package or our test harness, please consider citing \none of our related works.\n\n#### [Polynomial-time algorithms based on chain-based tree partitioning:](https://doi.org/10.1109/noms56928.2023.10154412)\n\nJ. Czentye, I. Pelle and B. Sonkoly,\n\"Cost-optimal Operation of Latency Constrained Serverless Applications: From Theory to Practice,\"\n_NOMS 2023-2023 IEEE/IFIP Network Operations and Management Symposium_, Miami, FL, USA, 2023, pp. 1-10,\ndoi: 10.1109/NOMS56928.2023.10154412.\n\n```bibtex\n@INPROCEEDINGS{Czentye2022noms,\n author = {J{\\'{a}}nos Czentye and Istv{\\'{a}}n Pelle and Bal{\\'{a}}zs Sonkoly},\n booktitle = {{NOMS 2023-2023 IEEE/IFIP Network Operations and Management Symposium}},\n title = {{Cost-optimal Operation of Latency Constrained Serverless Applications: From Theory to Practice}},\n publisher = {{IEEE}},\n year = {2023},\n month = may,\n pages = {1--10},\n doi = {10.1109/NOMS56928.2023.10154412}\n}\n```\n\n#### [Heuristic algorithm for dynamic (re)optimization control loop in edge-could environments:](https://doi.org/10.1109/jiot.2020.3042428)\n\nI. Pelle, J. Czentye, J. D\u00f3ka, A. Kern, B. P. Ger\u0151 and B. Sonkoly,\n\"Operating Latency Sensitive Applications on Public Serverless Edge Cloud Platforms,\"\nin _IEEE Internet of Things Journal_, vol. 8, no. 10, pp. 7954-7972, 15 May, 2021,\ndoi: 10.1109/JIOT.2020.3042428.\n\n```bibtex\n@ARTICLE{Pelle2021jiot,\n author = {Pelle, Istv{\\'{a}}n and Czentye, J{\\'{a}}nos and D{\\'{o}}ka, J{\\'{a}}nos and Kern, Andr{\\'{a}}s and Ger{\\H{o}}, Bal{\\'{a}}zs P. and Sonkoly, Bal{\\'{a}}zs},\n journal = {{IEEE Internet of Things Journal}},\n title = {{Operating Latency Sensitive Applications on Public Serverless Edge Cloud Platforms}},\n publisher = {Institute of Electrical and Electronics Engineers ({IEEE})},\n year = {2021},\n month = may,\n volume = {8},\n number = {10},\n pages = {7954--7972},\n doi = {10.1109/JIOT.2020.3042428}\n}\n```\n\n#### [Layout optimization for serverless applications over public clouds:](https://doi.org/10.1109/globecom38437.2019.9013988)\n\nJ. Czentye, I. Pelle, A. Kern, B. P. Gero, L. Toka and B. Sonkoly,\n\"Optimizing Latency Sensitive Applications for Amazon's Public Cloud Platform,\"\n_2019 IEEE Global Communications Conference (GLOBECOM)_, Waikoloa, HI, USA, 2019, pp. 1-7,\ndoi: 10.1109/GLOBECOM38437.2019.9013988.\n\n```bibtex\n@INPROCEEDINGS{Czentye2019globecom,\n author = {Czentye, Janos and Pelle, Istvan and Kern, Andras and Gero, Balazs Peter and Toka, Laszlo and Sonkoly, Balazs},\n booktitle = {{2019 IEEE Global Communications Conference (GLOBECOM)}},\n title = {{Optimizing Latency Sensitive Applications for Amazon's Public Cloud Platform}},\n publisher = {{IEEE}},\n year = {2019},\n month = dec,\n pages = {1--7},\n doi = {10.1109/GLOBECOM38437.2019.9013988}\n}\n```\n\n## License\n\nSLAMBUC is an open-source software licensed under [Apache 2.0](LICENSE).\n",
"bugtrack_url": null,
"license": "Apache 2.0",
"summary": "Serverless Layout Adaptation with Memory-Bounds and User Constraints",
"version": "0.2.1",
"project_urls": {
"Homepage": "https://github.com/hsnlab/SLAMBUC/wiki",
"Issue Tracker": "https://github.com/hsnlab/SLAMBUC/issues",
"Repository": "https://github.com/hsnlab/SLAMBUC"
},
"split_keywords": [
"cloud",
"serverless",
"ilp",
"dp",
"tree"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "9e3d21eff603e03c62fc18170ed1054e75c5d8039ae142810bec0721810bdd92",
"md5": "281a472b10b8779f7be9a42140a57bc0",
"sha256": "31d9f135a796c8b5717b7c3dea5fb8f04e1a28e5e91a582aafa9971d063b5990"
},
"downloads": -1,
"filename": "SLAMBUC-0.2.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "281a472b10b8779f7be9a42140a57bc0",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.10",
"size": 694883,
"upload_time": "2023-09-06T16:41:07",
"upload_time_iso_8601": "2023-09-06T16:41:07.597151Z",
"url": "https://files.pythonhosted.org/packages/9e/3d/21eff603e03c62fc18170ed1054e75c5d8039ae142810bec0721810bdd92/SLAMBUC-0.2.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "6eed3ad8c0488a2f79bec306e32e225b55b563301082fc12c784e56bdc9d5b79",
"md5": "a3e7f1cd5e8780b82d7207e9bedb7940",
"sha256": "bfff0817e2c0f982c2223d1f486e9e01815f1df503deb47997aa8b01ce4ac396"
},
"downloads": -1,
"filename": "SLAMBUC-0.2.1.tar.gz",
"has_sig": false,
"md5_digest": "a3e7f1cd5e8780b82d7207e9bedb7940",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.10",
"size": 637551,
"upload_time": "2023-09-06T16:41:10",
"upload_time_iso_8601": "2023-09-06T16:41:10.887363Z",
"url": "https://files.pythonhosted.org/packages/6e/ed/3ad8c0488a2f79bec306e32e225b55b563301082fc12c784e56bdc9d5b79/SLAMBUC-0.2.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2023-09-06 16:41:10",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "hsnlab",
"github_project": "SLAMBUC",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"requirements": [],
"lcname": "slambuc"
}
