Add example of tuning RocksDB on NNI

docs/en_US/TrialExample/RocksdbExamples.md

@@ -0,0 +1,97 @@
+# Tuning RocksDB on NNI
+
+## Overview
+
+[RocksDB](https://github.com/facebook/rocksdb) is a popular high performance embedded key-value database used in production systems at various web-scale enterprises including Facebook, Yahoo!, and LinkedIn.. It is a fork of [LevelDB](https://github.com/google/leveldb) by Facebook optimized to exploit many central processing unit (CPU) cores, and make efficient use of fast storage, such as solid-state drives (SSD), for input/output (I/O) bound workloads.
+
+The performance of RocksDB is highly contingent on its tuning. However, because of the complexity of its underlying technology and a large number of configurable parameters, a good configuration is sometimes hard to obtain. NNI can help to address this issue. NNI supports many kinds of tuning algorithms to search the best configuration of RocksDB, and support many kinds of environments like local machine, remote servers and cloud. 
+
+This example illustrates how to use NNI to search the best configuration of RocksDB for a `fillrandom` benchmark supported by a benchmark tool `db_bench`, which is an official benchmark tool provided by RocksDB itself. Therefore, before running this example, please make sure NNI is installed and [`db_bench`](https://github.com/facebook/rocksdb/wiki/Benchmarking-tools) is in your `PATH`. Please refer to [here](../Tutorial/QuickStart.md) for detailed information about installation and preparing of NNI environment, and [here](https://github.com/facebook/rocksdb/blob/master/INSTALL.md) for compiling RocksDB as well as `db_bench`.
+
+We also provide a simple script [`db_bench_installation.sh`](../../../examples/trials/systems/rocksdb-fillrandom/db_bench_installation.sh) helping to compile and install `db_bench` as well as its dependencies on Ubuntu. Installing RocksDB on other systems can follow the same procedure.
+
+*code directory: [`example/trials/systems/rocksdb-fillrandom`](../../../examples/trials/systems/rocksdb-fillrandom)*
+
+## Experiment setup
+
+There are mainly three steps to setup an experiment of tuning systems on NNI. Define search space with a `json` file, write a benchmark code, and start NNI experiment by passing a config file to NNI manager.
+
+### Search Space
+
+For simplicity, this example tunes three parameters, `write_buffer_size`, `min_write_buffer_num` and `level0_file_num_compaction_trigger`, for writing 16M keys with 20 Bytes of key size and 100 Bytes of value size randomly, based on writing operations per second (OPS). `write_buffer_size` sets the size of a single memtable. Once memtable exceeds this size, it is marked immutable and a new one is created. `min_write_buffer_num` is the minimum number of memtables to be merged before flushing to storage. Once the number of files in level 0 reaches `level0_file_num_compaction_trigger`, level 0 to level 1 compaction is triggered.
+
+In this example, the search space is specified by a `search_space.json` file as shown below. Detailed explanation of search space could be found [here](../Tutorial/SearchSpaceSpec.md).
+
+```json
+{
+    "write_buffer_size": {
+        "_type": "quniform",
+        "_value": [2097152, 16777216, 1048576]
+    },
+    "min_write_buffer_number_to_merge": {
+        "_type": "quniform",
+        "_value": [2, 16, 1]
+    },
+    "level0_file_num_compaction_trigger": {
+        "_type": "quniform",
+        "_value": [2, 16, 1]
+    }
+}
+```
+
+*code directory: [`example/trials/systems/rocksdb-fillrandom/search_space.json`](../../../examples/trials/systems/rocksdb-fillrandom/search_space.json)*
+
+### Benchmark code
+
+Benchmark code should receive a configuration from NNI manager, and report the corresponding benchmark result back. Following NNI APIs are designed for this purpose. In this example, writing operations per second (OPS) is used as a performance metric. Please refer to [here](Trials.md) for detailed information.
+
+* Use `nni.get_next_parameter()` to get next system configuration.
+* Use `nni.report_final_result(metric)` to report the benchmark result.
+
+*code directory: [`example/trials/systems/rocksdb-fillrandom/main.py`](../../../examples/trials/systems/rocksdb-fillrandom/main.py)*
+
+### Config file
+
+One could start a NNI experiment with a config file. A config file for NNI is a `yaml` file usually including experiment settings (`trialConcurrency`, `maxExecDuration`, `maxTrialNum`, `trial gpuNum`, etc.), platform settings (`trainingServicePlatform`, etc.), path settings (`searchSpacePath`, `trial codeDir`, etc.) and tuner settings (`tuner`, `tuner optimize_mode`, etc.). Please refer to [here](../Tutorial/QuickStart.md) for more information.
+
+Here is an example of tuning RocksDB with SMAC algorithm:
+
+*code directory: [`example/trials/systems/rocksdb-fillrandom/config_smac.yml`](../../../examples/trials/systems/rocksdb-fillrandom/config_smac.yml)*
+
+Here is an example of tuning RocksDB with TPE algorithm:
+
+*code directory: [`example/trials/systems/rocksdb-fillrandom/config_tpe.yml`](../../../examples/trials/systems/rocksdb-fillrandom/config_tpe.yml)*
+
+Other tuners can be easily adopted in the same way. Please refer to [here](../Tuner/BuiltinTuner.md) for more information.
+
+Finally, we could enter the example folder and start the experiment using following commands:
+
+```bash
+# tuning RocksDB with SMAC tuner
+nnictl create --config ./config_smac.yml
+# tuning RocksDB with TPE tuner
+nnictl create --config ./config_tpe.yml
+```
+
+## Experiment results
+
+We ran these two examples on the same machine with following details:
+
+* 16 * Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
+* 465 GB of rotational hard drive with ext4 file system
+* 128 GB of RAM
+* Kernel version: 4.15.0-58-generic
+* NNI version: v1.0-37-g1bd24577
+* RocksDB version: 6.4
+* RocksDB DEBUG_LEVEL: 0
+
+The detailed experiment results are shown in the below figure. Horizontal axis is sequential order of trials. Vertical axis is the metric, write OPS in this example. Blue dots represent trials for tuning RocksDB with SMAC tuner, and orange dots stand for trials for tuning RocksDB with TPE tuner. 
+
+![image](../../../examples/trials/systems/rocksdb-fillrandom/plot.png)
+
+Following table lists the best trials and corresponding parameters and metric obtained by the two tuners. Unsurprisingly, both of them found the same optimal configuration for `fillrandom` benchmark.
+
+| Tuner | Best trial | Best OPS | write_buffer_size | min_write_buffer_number_to_merge | level0_file_num_compaction_trigger |
+| :---: | :--------: | :------: | :---------------: | :------------------------------: | :--------------------------------: |
+| SMAC  | 255        | 779289   | 2097152           | 7.0                              | 7.0                                |
+| TPE   | 169        | 761456   | 2097152           | 7.0                              | 7.0                                |

docs/en_US/TrialExample/Trials.md

@@ -163,3 +163,4 @@ For more information, please refer to [HowToDebug](../Tutorial/HowToDebug.md)
 * [How to tune Scikit-learn on NNI](SklearnExamples.md)
 * [Automatic Model Architecture Search for Reading Comprehension.](SquadEvolutionExamples.md)
 * [Tuning GBDT on NNI](GbdtExample.md)
+* [Tuning RocksDB on NNI](RocksdbExamples.md)

docs/zh_CN/TrialExample/RocksdbExamples.md

@@ -0,0 +1,97 @@
+# 使用 NNI 调优 RocksDB
+
+## 概述
+
+[RocksDB](https://github.com/facebook/rocksdb) 是一种很受欢迎的高性能嵌入式键值数据库，被许多公司，如 Facebook， Yahoo! 和 LinkedIn 等，广泛应用于各种网络规模的产品中。它是 Facebook 在 [LevelDB](https://github.com/google/leveldb) 的基础上，通过充分利用多核心中央处理器和快速存储器（如固态硬盘）的特点，针对IO密集型应用优化而成的。
+
+RocksDB 的性能高度依赖于运行参数的调优。然而，由于其底层技术极为复杂，需要调整的参数过多，有时很难找到合适的运行参数。NNI 可以帮助数据库运维工程师解决这个问题。NNI 支持多种自动调参算法，并且支持运行于本地、远程和云端的各种负载。
+
+本例展示了如何使用 NNI 搜索 RocksDB 在 `fillrandom` 基准测试中的最佳运行参数，`fillrandom` 基准测试是 RocksDB 官方提供的基准测试工具 `db_bench` 所支持的一种基准测试，因此在运行本例之前请确保您已经安装了 NNI，并且 `db_bench` 在您的 `PATH` 路径中。关于如何安装和准备 NNI 环境，请参考[这里](../Tuner/BuiltinTuner.md)，关于如何编译 RocksDB 以及 `db_bench`，请参考[这里](https://github.com/facebook/rocksdb/blob/master/INSTALL.md)。
+
+我们还提供了一个简单的脚本 [`db_bench_installation.sh`](../../../examples/trials/systems/rocksdb-fillrandom/db_bench_installation.sh)，用来在 Ubuntu 系统上编译和安装 `db_bench` 和相关依赖。在其他系统中的安装也可以参考该脚本实现。
+
+*代码目录: [`example/trials/systems/rocksdb-fillrandom`](../../../examples/trials/systems/rocksdb-fillrandom)*
+
+## 实验配置
+
+使用 NNI 进行调优系统主要有三个步骤，分别是，使用一个 `json` 文件定义搜索空间；准备一个基准测试程序；和一个用来启动 NNI 实验的配置文件。
+
+### 搜索空间
+
+简便起见，本例基于 Rocks_DB 每秒的写入操作数（Operations Per Second, OPS），在随机写入 16M 个键长为 20 字节值长为 100 字节的键值对的情况下，对三个系统运行参数，`write_buffer_size`，`min_write_buffer_num` 和 `level0_file_num_compaction_trigger`，进行了调优。`write_buffer_size` 控制了单个 memtable 的大小。在写入过程中，当 memtable 的大小超过了 `write_buffer_size` 指定的数值，该 memtable 将会被标记为不可变，并创建一个新的 memtable。`min_write_buffer_num` 是在写入（flush）磁盘之前需要合并（merge）的 memtable 的最小数量。一旦 level 0 中的文件数量超过了 `level0_file_num_compaction_trigger` 所指定的数，level 0 向 level 1 的压缩（compaction）将会被触发。
+
+搜索空间由如下所示的文件 `search_space.json` 指定。更多关于搜索空间的解释请参考[这里](../Tutorial/SearchSpaceSpec.md)。
+
+```json
+{
+    "write_buffer_size": {
+        "_type": "quniform",
+        "_value": [2097152, 16777216, 1048576]
+    },
+    "min_write_buffer_number_to_merge": {
+        "_type": "quniform",
+        "_value": [2, 16, 1]
+    },
+    "level0_file_num_compaction_trigger": {
+        "_type": "quniform",
+        "_value": [2, 16, 1]
+    }
+}
+```
+
+*代码目录: [`example/trials/systems/rocksdb-fillrandom/search_space.json`](../../../examples/trials/systems/rocksdb-fillrandom/search_space.json)*
+
+### 基准测试
+
+基准测试程序需要从 NNI manager 接收一个运行参数，并在运行基准测试以后向 NNI manager 汇报基准测试结果。NNI 提供了下面两个 APIs 来完成这些任务。更多关于 NNI trials 的信息请参考[这里](Trials.md)。
+
+* 使用 `nni.get_next_parameter()` 从 NNI manager 得到需要测试的系统运行参数。
+* 使用 `nni.report_final_result(metric)` 向 NNI manager 汇报基准测试的结果。
+
+*代码目录: [`example/trials/systems/rocksdb-fillrandom/main.py`](../../../examples/trials/systems/rocksdb-fillrandom/main.py)*
+
+### 配置文件
+
+NNI 实验可以通过配置文件来启动。通常而言，NNI 配置文件是一个 `yaml` 文件，通常包含实验设置（`trialConcurrency`，`maxExecDuration`，`maxTrialNum`，`trial gpuNum` 等），运行平台设置（`trainingServicePlatform` 等），路径设置（`searchSpacePath`，`trial codeDir` 等）和 调参器设置（`tuner`，`tuner optimize_mode` 等）。更多关于 NNI 配置文件的信息请参考[这里](../Tutorial/QuickStart.md)。
+
+下面是使用 SMAC 算法调优 RocksDB 配置文件的例子：
+
+*代码目录: [`example/trials/systems/rocksdb-fillrandom/config_smac.yml`](../../../examples/trials/systems/rocksdb-fillrandom/config_smac.yml)*
+
+下面是使用 TPE 算法调优 RocksDB 配置文件的例子：
+
+*代码目录: [`example/trials/systems/rocksdb-fillrandom/config_tpe.yml`](../../../examples/trials/systems/rocksdb-fillrandom/config_tpe.yml)*
+
+其他的调参器可以使用同样的方式应用，更多关于调参器的信息请参考[这里](../Tuner/BuiltinTuner.md)。
+
+最后，我们可以进入本例的文件夹内，用下面的命令启动实验：
+
+```bash
+# tuning RocksDB with SMAC tuner
+nnictl create --config ./config_smac.yml
+# tuning RocksDB with TPE tuner
+nnictl create --config ./config_tpe.yml
+```
+
+## 实验结果
+
+我们在同一台机器上运行了这两个实验，相关信息如下：
+
+* 16 * Intel(R) Xeon(R) CPU E5-2650 v2 @ 2.60GHz
+* 465 GB of rotational hard drive with ext4 file system
+* 128 GB of RAM
+* Kernel version: 4.15.0-58-generic
+* NNI version: v1.0-37-g1bd24577
+* RocksDB version: 6.4
+* RocksDB DEBUG_LEVEL: 0
+
+具体的实验结果如下图所示。横轴是基准测试的顺序，纵轴是基准测试得到的结果，在本例中是每秒钟写操作的次数。蓝色的圆点代表用 SMAC 调优 RocksDB 得到的基准测试结果，而橘黄色的圆点表示用 TPE 调优得到的基准测试结果。
+
+![image](../../../examples/trials/systems/rocksdb-fillrandom/plot.png)
+
+下面的表格列出了使用两种调参器得到的最好的基准测试结果及相对应的参数。毫不意外，使用这两种调参器在 `fillrandom` 基准测试中搜索得到了相同的最优参数。
+
+| Tuner | Best trial | Best OPS | write_buffer_size | min_write_buffer_number_to_merge | level0_file_num_compaction_trigger |
+| :---: | :--------: | :------: | :---------------: | :------------------------------: | :--------------------------------: |
+| SMAC  | 255        | 779289   | 2097152           | 7.0                              | 7.0                                |
+| TPE   | 169        | 761456   | 2097152           | 7.0                              | 7.0                                |

docs/zh_CN/TrialExample/Trials.md

@@ -168,4 +168,5 @@ echo $? `date +%s%3N` >/home/user_name/nni/experiments/$experiment_id$/trials/$t
 * [为 CIFAR 10 分类找到最佳的 optimizer](Cifar10Examples.md)
 * [如何在 NNI 调优 SciKit-learn 的参数](SklearnExamples.md)
 * [在阅读理解上使用自动模型架构搜索。](SquadEvolutionExamples.md)
-* [如何在 NNI 上调优 GBDT](GbdtExample.md)
+* [如何在 NNI 上调优 GBDT](GbdtExample.md)
+* [如何在 NNI 上调优 RocksDB](RocksdbExamples.md)

examples/trials/systems/rocksdb-fillrandom/config_smac.yml

@@ -0,0 +1,21 @@
+authorName: default
+experimentName: auto_rocksdb_SMAC
+trialConcurrency: 1
+maxExecDuration: 12h
+maxTrialNum: 256
+#choice: local, remote, pai
+trainingServicePlatform: local
+searchSpacePath: search_space.json
+#choice: true, false
+useAnnotation: false
+tuner:
+  #choice: TPE, Random, Anneal, Evolution, BatchTuner, MetisTuner
+  #SMAC (SMAC should be installed through nnictl)
+  builtinTunerName: SMAC
+  classArgs:
+    #choice: maximize, minimize
+    optimize_mode: maximize
+trial:
+  command: python3 main.py
+  codeDir: .
+  gpuNum: 0

examples/trials/systems/rocksdb-fillrandom/config_tpe.yml

@@ -0,0 +1,21 @@
+authorName: default
+experimentName: auto_rocksdb_TPE
+trialConcurrency: 1
+maxExecDuration: 12h
+maxTrialNum: 256
+#choice: local, remote, pai
+trainingServicePlatform: local
+searchSpacePath: search_space.json
+#choice: true, false
+useAnnotation: false
+tuner:
+  #choice: TPE, Random, Anneal, Evolution, BatchTuner, MetisTuner
+  #SMAC (SMAC should be installed through nnictl)
+  builtinTunerName: TPE
+  classArgs:
+    #choice: maximize, minimize
+    optimize_mode: maximize
+trial:
+  command: python3 main.py
+  codeDir: .
+  gpuNum: 0

examples/trials/systems/rocksdb-fillrandom/db_bench_installation.sh

@@ -0,0 +1,36 @@
+#!/bin/bash
+# Install db_bench and its dependencies on Ubuntu
+
+pushd $PWD 1>/dev/null
+
+# install snappy
+echo "****************** Installing snappy *******************"
+sudo apt-get install libsnappy-dev -y
+
+# install gflag
+echo "****************** Installing gflag ********************"
+cd /tmp
+git clone https://github.com/gflags/gflags.git
+cd gflags
+git checkout v2.0
+./configure && make && sudo make install
+
+# install rocksdb
+echo "****************** Installing rocksdb ******************"
+cd /tmp
+git clone https://github.com/facebook/rocksdb.git
+cd rocksdb
+CPATH=/usr/local/include LIBRARY_PATH=/usr/local/lib DEBUG_LEVEL=0 make db_bench -j7
+
+DIR=$HOME/.local/bin/
+if [[ ! -e $DIR ]]; then
+    mkdir $dir
+elif [[ ! -d $DIR ]]; then
+    echo "$DIR already exists but is not a directory" 1>&2
+    exit
+fi
+mv db_bench $HOME/.local/bin &&
+echo "Successfully installed rocksed in "$DIR" !" &&
+echo "Please add "$DIR" to your PATH for runing this example."
+
+popd 1>/dev/null

examples/trials/systems/rocksdb-fillrandom/main.py

@@ -0,0 +1,96 @@
+# Copyright (c) Microsoft Corporation
+# All rights reserved.
+#
+# MIT License
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated
+# documentation files (the "Software"), to deal in the Software without restriction, including without limitation
+# the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and
+# to permit persons to whom the Software is furnished to do so, subject to the following conditions:
+# The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED *AS IS*, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING
+# BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
+# DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+import nni
+import subprocess
+import logging
+
+LOG = logging.getLogger('rocksdb-fillrandom')
+
+
+def run(**parameters):
+    '''Run rocksdb benchmark and return throughput'''
+    bench_type = parameters['benchmarks']
+    # recover args
+    args = ["--{}={}".format(k, v) for k, v in parameters.items()]
+    # subprocess communicate
+    process = subprocess.Popen(['db_bench'] + args, stdout=subprocess.PIPE)
+    out, err = process.communicate()
+    # split into lines
+    lines = out.decode("utf8").splitlines()
+
+    match_lines = []
+    for line in lines:
+        # find the line with matched str
+        if bench_type not in line:
+            continue
+        else:
+            match_lines.append(line)
+            break
+
+    results = {}
+    for line in match_lines:
+        key, _, value = line.partition(":")
+        key = key.strip()
+        value = value.split("op")[1]
+        results[key] = float(value)
+
+    return results[bench_type]
+
+
+def generate_params(received_params):
+    '''generate parameters based on received parameters'''
+    params = {
+        "benchmarks": "fillrandom",
+        "threads": 1,
+        "key_size": 20,
+        "value_size": 100,
+        "num": 13107200,
+        "db": "/tmp/rockdb",
+        "disable_wal": 1,
+        "max_background_flushes": 1,
+        "max_background_compactions": 4,
+        "write_buffer_size": 67108864,
+        "max_write_buffer_number": 16,
+        "min_write_buffer_number_to_merge": 2,
+        "level0_file_num_compaction_trigger": 2,
+        "max_bytes_for_level_base": 268435456,
+        "max_bytes_for_level_multiplier": 10,
+        "target_file_size_base": 33554432,
+        "target_file_size_multiplier": 1
+    }
+
+    for k, v in received_params.items():
+        params[k] = int(v)
+
+    return params
+
+
+if __name__ == "__main__":
+    try:
+        # get parameters from tuner
+        RECEIVED_PARAMS = nni.get_next_parameter()
+        LOG.debug(RECEIVED_PARAMS)
+        PARAMS = generate_params(RECEIVED_PARAMS)
+        LOG.debug(PARAMS)
+        # run benchmark
+        throughput = run(**PARAMS)
+        # report throughput to nni
+        nni.report_final_result(throughput)
+    except Exception as exception:
+        LOG.exception(exception)
+        raise