AresDB is a GPU-powered real-time analytics storage and query engine. It features low query latency, high data freshness and highly efficient in-memory and on disk storage management. Please see AresDB's features, architecture design described in the Uber Engineering Blog.

This repo contains the source code of AresDB and debug UI.

AresDB requires the CUDA Toolkit. Please ensure you read, acknowledge, and accept the CUDA End User License Agreement.

To get AresDB:

git clone --recursive https://github.com/uber/aresdb.git $GOPATH/src/github.com/uber/aresdb

AresDB needs NVIDIA driver version >= 390.48 and CUDA version 9.1.

Run the following to make sure the following environment variables are correctly set:

export PATH=/path/to/cuda/bin:${PATH}
export LD_LIBRARY_PATH=/path/to/cuda/lib64:/path/to/aresdb/lib:${LD_LIBRARY_PATH}
export PKG_CONFIG_PATH=${LD_LIBRARY_PATH}/pkgconfig:${PKG_CONFIG_PATH}

Building and running AresDB requires:

• golang 1.11+
• C++ compiler that support c++14
• cmake 3.12+
• nvcc version 9.1

Run following commands to generate makefile:

cmake -DQUERY_MODE=DEVICE .

Alternatively, if you want to run the query in CPU mode, run following commands:

cmake -DQUERY_MODE=HOST .

AresDB is written in C++ (query engine) and Golang (mem store, disk store and other query components). Because of this, we break testing into two parts:

We use Ginkgo as the test framework for running the Golang unit test and coverage. Install Ginkgo first and run

make test-golang

We use google-test as the test framework to test C++ code. Install google-test and set the environment variable, GTEST_ROOT, to the installed location.

After you have installed properly, run

make test-cuda

The following command will start an AresDB server locally. You can start to query the server using a curl command or swagger page.

make run_server

Please read the Docker page.

Interested in learning more about AresDB? Read the blog post

Apache 2.0 License, please see LICENSE for details.