diff options
Diffstat (limited to 'site/docs')
| -rw-r--r-- | site/docs/advanced-guides/_category_.json | 7 | ||||
| -rw-r--r-- | site/docs/advanced-guides/architecture.md | 26 | ||||
| -rw-r--r-- | site/docs/advanced-guides/deploy.md | 84 | ||||
| -rw-r--r-- | site/docs/advanced-guides/img/component-diagram.png | bin | 39965 -> 0 bytes | |||
| -rw-r--r-- | site/docs/advanced-guides/toolchain.md | 74 | ||||
| -rw-r--r-- | site/docs/documentation/Output.md | 5 | ||||
| -rw-r--r-- | site/docs/getting-started/3-whats-next.md | 1 | ||||
| -rw-r--r-- | site/docs/tutorials/cloud-capacity-planning.mdx | 277 |
8 files changed, 3 insertions, 471 deletions
diff --git a/site/docs/advanced-guides/_category_.json b/site/docs/advanced-guides/_category_.json deleted file mode 100644 index a74f4f42..00000000 --- a/site/docs/advanced-guides/_category_.json +++ /dev/null @@ -1,7 +0,0 @@ -{ - "label": "Advanced Guides", - "position": 4, - "link": { - "type": "generated-index" - } -} diff --git a/site/docs/advanced-guides/architecture.md b/site/docs/advanced-guides/architecture.md deleted file mode 100644 index 2a65a6c6..00000000 --- a/site/docs/advanced-guides/architecture.md +++ /dev/null @@ -1,26 +0,0 @@ ---- -sidebar_position: 2 ---- - -# Architecture - -OpenDC consists of four components: a Kotlin simulator, a SQL database, a Quarkus-based -[API](https://github.com/atlarge-research/opendc/tree/master/opendc-web/opendc-web-api), and a -React.js [frontend](https://github.com/atlarge-research/opendc/tree/master/opendc-web/opendc-web-api). - - - -On the frontend, users can construct a topology by specifying a datacenter's rooms, racks and machines, and create -scenarios to see how a workload trace runs on that topology. The frontend communicates with the web server via a REST -API over HTTP. - -The (Swagger/OpenAPI compliant) API spec specifies what requests the frontend can make to the web server. To view this -specification, go to the [Swagger Editor](https://editor.swagger.io/) and paste in -our [API spec](https://api.opendc.org/q/openapi). - -The web server receives API requests and processes them in the database. When the frontend requests to run a new -scenario, the web server adds it to the `scenarios` collection in the database and sets its `state` as `PENDING`. - -The simulator monitors the database for `PENDING` scenarios, and simulates them as they are submitted. The results of -the simulations are processed and aggregated in memory. Afterwards, the aggregated summary is written to the database, -which the frontend can then again retrieve via the web server. diff --git a/site/docs/advanced-guides/deploy.md b/site/docs/advanced-guides/deploy.md deleted file mode 100644 index 2ee69c07..00000000 --- a/site/docs/advanced-guides/deploy.md +++ /dev/null @@ -1,84 +0,0 @@ ---- -sidebar_position: 3 ---- - -# Deploying OpenDC -This document explains how you can deploy a multi-tenant instance of OpenDC using Docker. - -## Contents - -1. [Setting up Auth0](#setting-up-auth0) -1. [Installing Docker](#installing-docker) -1. [Running OpenDC from source](#running-opendc-from-source) - -## Setting up Auth0 - -OpenDC uses [Auth0](https://auth0.com) as Identity Provider so that OpenDC does not have to manage user data itself, -which greatly simplifies our frontend and backend implementation. We have chosen to use Auth0 as it is a well-known -Identity Provider with good software support and a free tier for users to experiment with. - -To deploy OpenDC yourself, you need to have an [Auth0 tenant](https://auth0.com/docs/get-started/learn-the-basics) and -create: - -1. **An API** - You need to define the OpenDC API server in Auth0. Please refer to the [following guide](https://auth0.com/docs/quickstart/backend/python/01-authorization#create-an-api) - on how to define an API in Auth0. - - Remember the identifier you created the API with, as we need it in the next steps (as `OPENDC_AUTH0_AUDIENCE`). -2. **A Single Page Application (SPA)** - You need to define the OpenDC frontend application in Auth0. Please see the [following guide](https://auth0.com/docs/quickstart/spa/react#configure-auth0) - on how you can define an SPA in Auth0. Make sure you have added the necessary URLs to the _Allowed Callback URLs_: - for a local deployment, you should add at least `http://localhost:3000, http://localhost:8080`. - - Once your application has been created, you should have a _Domain_ and _Client ID_ which we need to pass to the - frontend application (as `OPENDC_AUTH0_DOMAIN` and `OPENDC_AUTH0_CLIENT_ID` respectively). - - -## Installing Docker - -OpenDC uses [Docker](https://www.docker.com/) and [Docker Compose](https://docs.docker.com/compose/) to orchestrate the -deployment of the software stack. Please refer to [Docker Desktop](https://www.docker.com/products/docker-desktop) for -instructions on how install Docker on your machine. - -## Running OpenDC from source - -To build and run the full OpenDC stack locally on Linux or Mac, you first need to clone the project: - -```bash -git clone https://github.com/atlarge-research/opendc.git - -# Enter the directory -cd opendc/ -``` - -In the directory you just entered, you need to set up a set of environment variables. To do this, create a file -called `.env` in the `opendc` folder. In this file, replace `your-auth0-*` with the Auth0 details you got from the first -step. For a standard setup, you can leave the other settings as-is. - -```.env -OPENDC_DB_USERNAME=opendc -OPENDC_DB_PASSWORD=opendcpassword -OPENDC_AUTH0_DOMAIN=your-auth0-domain -OPENDC_AUTH0_CLIENT_ID=your-auth0-client-id -OPENDC_AUTH0_AUDIENCE=your-auth0-api-identifier -OPENDC_API_BASE_URL=http://web -``` - -We provide a set of default traces for you to experiment with. If you want to add others, place them in the `traces` -directory and add entries to the database (see also [the SQL init script](https://github.com/atlarge-research/opendc/tree/master/opendc-web/opendc-web-server/src/main/resources/db/migration/V1.0.0__core.sql)) - -If you plan to deploy publicly, please also tweak the other settings. In that case, also check the `docker-compose.yml` -and `docker-compose.prod.yml` for further instructions. - -Now, start the server: - -```bash -# Build the Docker image -docker-compose build - -# Start the containers -docker-compose up -``` - -Wait a few seconds and open `http://localhost:8080` in your browser to use OpenDC. We recommend Google Chrome for the -best user experience. diff --git a/site/docs/advanced-guides/img/component-diagram.png b/site/docs/advanced-guides/img/component-diagram.png Binary files differdeleted file mode 100644 index 312ca72a..00000000 --- a/site/docs/advanced-guides/img/component-diagram.png +++ /dev/null diff --git a/site/docs/advanced-guides/toolchain.md b/site/docs/advanced-guides/toolchain.md deleted file mode 100644 index 1673d974..00000000 --- a/site/docs/advanced-guides/toolchain.md +++ /dev/null @@ -1,74 +0,0 @@ ---- -sidebar_position: 1 ---- - -# Toolchain Setup - -The OpenDC simulator is built using the [Kotlin](https://kotlinlang.org/) language. This is a JVM-based language that -should appear familiar to programmers knowledgeable in Java or Scala. For a short interactive introduction to Kotlin, -the [Learn Kotlin By Example](https://play.kotlinlang.org/byExample/overview) docs are a great place to start. - -For the build and dependency toolchain, we use [Gradle](https://gradle.org/). You will likely not need to change the -Gradle build configurations of components, but you will use Gradle to execute builds and tests on the codebase. - -Follow the steps below to get it all set up! - -## Contents - -1. [Installing Java](#1-installing-java) -2. [Building and Developing](#2-building-and-developing) -3. [Setup with IntelliJ IDEA](#21-setup-with-intellij-idea) -4. [Setup with Command Line](#22-setup-with-command-line) - -## 1. Installing Java - -OpenDC requires a Java installation of version 17 or higher. Make sure to install -the [JDK](https://www.oracle.com/technetwork/java/javase/downloads/index.html), not only the JRE (the JDK also includes -a JRE). - -## 2. Building and Developing - -With Java installed, we're ready to set up the development environment on your machine. You can either use a visual IDE -or work from a command line shell. We outline both approaches below, feel free to choose which you are most comfortable -with. If in doubt which one to choose, we suggest going with the first one. - -## 2.1 Setup with IntelliJ IDEA - -We suggest using [IntelliJ IDEA](https://www.jetbrains.com/idea/) as development environment. Once you have installed -any version of this IDE on your machine, choose "Get from Version Control" in the new project dialogue. -Enter `https://github.com/atlarge-research/opendc` as URL and submit your credentials when asked. -Open the project once it's ready fetching the codebase, and let it set up with the defaults (IntelliJ will recognize -that this is a Gradle codebase). - -You will now be prompted in a dialogue to enable auto-import for Gradle, which we suggest you do. Wait for any progress -bars in the lower bar to disappear and then look for the Gradle simHyperVisorContext menu on the right-hand side. In it, go -to `opendc > Tasks > verification > test`. This will build the codebase and run checks to verify that tests -pass. If you get a `BUILD SUCCESSFUL` message, you're ready to go to the [next section](architecture)! - -## 2.2 Setup with Command Line - -First, clone the repository with the following command: - -```shell script -git clone https://github.com/atlarge-research/opendc -``` - -And enter the directory: - -```shell script -cd opendc -``` - -If on Windows, run the batch file included in the root, as follows: - -```commandline -gradlew.bat test -``` - -If on Linux/macOS, run the shell script included in the root, as follows: - -```shell script -./gradlew test -``` - -If the build is successful, you are ready to go to the next section! diff --git a/site/docs/documentation/Output.md b/site/docs/documentation/Output.md index 3f9eb3d5..339ac615 100644 --- a/site/docs/documentation/Output.md +++ b/site/docs/documentation/Output.md @@ -1,6 +1,7 @@ -Running OpenDC results in three output files. The first file ([Server](#server)) contains metrics related to the jobs being executed. -The second file ([Host](#host)) contains all metrics related to the hosts on which jobs can be executed. Finally, the third file ([Service](#service)) +Running OpenDC results in three output files. The first file ([Task](#task)) contains metrics related to the jobs being executed. +The second file ([Host](#host)) contains all metrics related to the hosts on which jobs can be executed. The third file ([Power](#power)) +contains all metrics related to the power sources that power the hosts. Finally, the third file ([Service](#service)) contains metrics describing the overall performance. An experiment in OpenDC has ### Task diff --git a/site/docs/getting-started/3-whats-next.md b/site/docs/getting-started/3-whats-next.md index 7c021119..e613eba7 100644 --- a/site/docs/getting-started/3-whats-next.md +++ b/site/docs/getting-started/3-whats-next.md @@ -7,6 +7,5 @@ description: How to supercharge your designs and experiments with OpenDC. Congratulations! You have just learned how to design and experiment with a (virtual) datacenter in OpenDC. What's next? - Follow one of the [tutorials](/docs/category/tutorials) using OpenDC. -- Check the [advanced guides](/docs/category/advanced-guides) for more complex material. - Read about [existing work using OpenDC](/community/research). - Get involved in the [OpenDC Community](/community/support). diff --git a/site/docs/tutorials/cloud-capacity-planning.mdx b/site/docs/tutorials/cloud-capacity-planning.mdx deleted file mode 100644 index df9cb566..00000000 --- a/site/docs/tutorials/cloud-capacity-planning.mdx +++ /dev/null @@ -1,277 +0,0 @@ ---- -sidebar_position: 1 -title: Cloud Capacity Planning -hide_title: true -sidebar_label: Cloud Capacity Planning -description: Cloud Capacity Planning ---- - -# Cloud Capacity Planning Tutorial - -Using OpenDC to plan and design cloud datacenters. - -:::info Learning goal - -By doing this assignment, you will learn more about the basic concepts of datacenters and cloud computing systems, as -well as how such systems are planned and designed. -::: - -## Preamble - -Datacenter infrastructure is important in today’s digital society. Stakeholders across industry, government, and -academia employ a vast and diverse array of cloud services hosted by datacenter infrastructure, and expect services to -be reliable, high speed, and low cost. In turn, datacenter operators must maintain efficient operation at unprecedented -scale. - -To keep up with growing demand and increasing complexity, architects of datacenters must address complex challenges in -distributed systems, software engineering and performance engineering. One of these challenges is efficient utilization -of resources in datacenters, which is only 6-12% industry-wide despite the fact that it is inconvenient for datacenter -operators to keep much of their infrastructure idle, due to resulting high energy consumption and thus unnecessary -costs. - -It is often quite difficult to implement optimizations or other changes in datacenters. Datacenter operators tend to be -conservative in adopting such changes in fear of failure or misbehaving systems. Furthermore, testing changes at the -scale of modern datacenter infrastructure in a real-world setting is prohibitively expensive and hard to reproduce, -notwithstanding environmental concerns. - -A more viable alternative is the use of datacenter simulators such as OpenDC or CloudSim. These tools model datacenter -infrastructure at a good accuracy and allow us to test changes in a controllable and repeatable environment. - -In this tutorial, we will use the OpenDC datacenter simulator to experiment with datacenters and demonstrate the -process of designing and optimizing datacenters using simulation. - -## What is OpenDC - -OpenDC is an open source platform for datacenter simulation developed by AtLarge Research. The purpose of OpenDC is -twofold: we aim to both enable cloud computing education and support research into datacenters. -An example of the former is this tutorial, and examples of the latter include the numerous BSc and MSc research projects -that are using OpenDC to run experiments and perform research. - -:::caution - -OpenDC is still an experimental tool. Your data may get lost, overwritten, or otherwise become unavailable. Sorry for -the inconvenience. - -::: - -If you are not familiar with the OpenDC web interface, please follow the [Getting Started](/docs/category/getting-started) -guide to get an understanding of the main concepts of OpenDC and how to design a datacenter. - -:::note Action - -Set up a project on the OpenDC website. - -::: - - -## Assignment - -Acme Inc. is a small datacenter operator in the Netherlands. They are currently in the process of acquiring a new client -and closing a deal where Acme will migrate all of the client’s business-critical workloads from internal machines to -Acme’s datacenters. With this deal, the client aims to outsource the maintenance of their digital infrastructure, but in -turn expects reliable and efficient operation from Acme. - -To demonstrate that Acme is capable of this task, it has started a pilot project with the client where Acme will migrate -already a small subset of the client’s workloads. You are an engineer at Acme. and have been tasked with the design and -procurement of the datacenter infrastructure required for this pilot project. - -To guide your design, the client has provided a workload trace of their business-critical workloads, which consist of -the historical runtime behavior of 50 virtual machines over time. These virtual machines differ in resource -requirements (e.g. number of vCPUs or memory) and in resource consumption over time. We can use OpenDC to simulate this -workload trace and validate your datacenter design. - -The assignment is divided into four parts: -1. Analyzing the requirements to estimate what resources are needed. -2. Building your design in OpenDC -3. Validating your design in OpenDC -4. Optimizing your design in OpenDC - -Make notes of your thoughts on the following assignments & questions and discuss with your partner(s). - -## Analyze the Requirements - -The first step of the assignment is to analyze the requirements of the client in order to come up with a reasonable -estimation of the datacenter infrastructure needed. This estimation will become our initial design which we will build -and validate in OpenDC. - -Since the client has provided a workload trace representative of the workload that will eventually be running in the -datacenter, we can use it to guide our design. In [Figure 1](#resource-distribution), the requested memory and vCPUs are -depicted for the virtual machines in the workload trace. - -:::note Action - -Determine the total amount of vCPUs and memory required in the trace. - -::: - -<figure className="figure" id="resource-distribution"> - <img src={require("./img/resource-distribution.png").default} alt="Resource requirements for the workload" /> - <figcaption> - Requested number of vCPUs and memory (in GB) by the - virtual machines in the workload. The left figure shows the number of virtual machines that have requested 1, 2, 4 or 8 - vCPUs. The right figure shows the amount of memory requested compared to the number of vCPUs in the virtual machine. - </figcaption> -</figure> - -Based on this information, we could choose to purchase a new machine for every virtual machine in the workload trace. -Such a design will most certainly be able to handle the workload. At the same time, it is much more expensive and -probably unnecessary. - -In [Figure 2](#cpu-usage), the CPU Usage (in MHz) of the virtual machines in the workload is depicted over time. Observe that the -median CPU usage of the virtual machines over the whole trace is approximately 100 MHz. This means that a 2-core -processor with a base clock 3500 MHz would have utilization of only 1.4% (`100 MHz / (3500 MHz x 2)`) for such a median -workload. - -<figure className="figure" id="cpu-usage"> - <img src={require("./img/cpu-usage.png").default} alt="CPU usage over time for the workload" /> - <figcaption>CPU Usage of the virtual machines in the workload over time.</figcaption> -</figure> - -Instead, we could try to fit multiple virtual machines onto a single machine. For instance, the 2-core processor -mentioned before is able to handle 70 virtual machines, each running at 100 MHz (`(3500 MHz x 2) / 100 MHz`), ignoring -virtualization overhead and memory requirements. - -:::note Action - -Make a rough estimate of the number of physical cores required to host the vCPUs in the workload trace. - -::: - -Now that we have an indication of the number of physical cores we need to have, we can start to compose the servers in -our datacenter. See **Table 1 and 2** for the equipment list you can choose from. Don’t forget to put enough memory in your -servers, or otherwise you risk that not all virtual machines will fit on the servers in your datacenter. - -| Processor | Intel® Xeon® E-2224G | Intel® Xeon® E-2244G | Intel® Xeon® E-2246G | -|----------------------------------|----------------------|----------------------|----------------------| -| Base clock (in MHz) | 3500 | 3800 | 3600 | -| Core count | 4 | 8 | 12 | -| Average power consumption (in W) | 71 | 71 | 80 | - -**Table 1:** Processor options for your datacenter - - -| Memory module | Crucial MTA9ASF2G72PZ-3G2E | Crucial MTA18ASF4G72PDZ-3G2E1 | -|----------------------------------|----------------------------|-------------------------------| -| Size (in GB) | 16 | 32 | -| Speed (in MHZ) | 3200 | 3200 | -**Table 2:** Memory options for your datacenter - - -:::note Action - -Create a plan detailing the servers you want to have in your datacenter and what resources (e.g. processor or memory) -they should contain. For instance, such a plan could look like: - -1. 8x Server (2x Intel® Xeon® E-2244G, 4x Crucial MTA18ASF4G72PDZ-3G2E1) - -::: - -:::tip Hint - -Budget more capacity than your initial estimates to prevent your datacenter from running at a very high -utilization. Think about how your datacenter would handle a machine failure, will you still have enough capacity left? - -::: - -## Build the datacenter - -Based on the plan we devised in the previous section, we will now construct a (virtual) datacenter in OpenDC. If you -have not yet used the OpenDC web interface to design a datacenter, please read [Getting Started](/docs/category/getting-started) -guide to get an understanding of the main concepts of OpenDC and how to design a datacenter. - -:::note Action - -Implement your plan in the OpenDC web interface. - -::: - -## Validate your design - -We are now at a stage where we can validate whether the datacenter we have just designed and built in OpenDC is suitable -for the workload of the client. We will use OpenDC to simulate the workload in our datacenter and keep track of several -metrics to ensure efficient and reliable operation. - -One of our concerns is that our datacenter does not have enough computing power to deal with the client’s -business-critical workload, leading to degraded performance and consequently an unhappy client. - -A metric that gives us an insight in performance degradation is the Overcommitted CPU Cycles, which represents the -number of CPU cycles that a virtual machine wanted to run, but could not due to the host machine not having enough -computing capacity at that moment. To keep track of this metric during simulation, we create a new portfolio by clicking -the ‘+’ next to “Portfolio” in the left sidebar and select the metrics of interest. - -:::note Action - -Add a new portfolio and select at least the following targets: -1. Overcommitted CPU Cycles -2. Granted CPU Cycles -3. Requested CPU Cycles -4. Maximum Number VMs Finished - -::: - -We will now try to simulate the client’s workload trace (called _Bitbrains (Sample)_ in OpenDC). By clicking on ‘New -Scenario’ below your created portfolio, we can create a base scenario which will represent our baseline datacenter -design which we will compare against future improvements. - -:::note Action - -Add a base scenario to your new portfolio and select as trace _Bitbrains (Sample)_. - -::: - -By creating a new scenario, you will schedule a simulation of your datacenter design that will run on one of the OpenDC -simulation servers. Press the Play button next to your portfolio to see the results of the simulations. If you have -chosen the _Bitbrains (Sample)_ trace, the results should usually appear within one minute or less depending on the queue -size. In case they do not appear within a reasonable timeframe, please contact the instructors. - -You can now see how your design has performed. Check whether all virtual machines have finished and whether the -_Overcommitted CPU Cycles_ metric is not too high. Try to aim for anything below 1 bn cycles. In the next section, we’ll -try to further optimize our design. For now, think of an explanation for the performance of your design. - -## Optimize your design - -Finally, let’s try to optimize your design so that it meets the requirements of the client and is beneficial for your -employer as well. In particular, your company is interested in the follow goals: - -1. Reducing _Overcommitted CPU Cycles_ to a minimum for reliability. -2. Reducing _Total Power Consumption_ to a minimum to save energy costs. - -:::note Action - -Add a new portfolio and select at least the following targets: -1. Overcommitted CPU Cycles -2. Granted CPU Cycles -3. Requested CPU Cycles -4. Total Power Consumption - -Then, add a base scenario to your new portfolio and select as trace _Bitbrains (Sample)_. - -::: - -Try to think of ways in which you can reduce both _Overcommitted CPU Cycles_ and _Total Power Consumption_. Create a new -topology based on your initial topology and apply the changes you have come up with. In this way, you can easily compare -the performance of different topologies in different scenarios. Note that the choice of scheduler might also influence -your results. - -:::tip Hint - -The choice of scheduler (and thus the placement of VMs) might also influence your results. - -::: - - -:::note Action - -1. Create a new topology based on your existing topology. -2. Add a new scenario to your created portfolio and select your newly created topology. -3. Compare the results against the base scenario. - -::: - -Repeat this approach until you are satisfied with your design. - -## Epilogue - -In this tutorial, you should have learned briefly about what datacenters are, and the process of designing and -optimizing a datacenter yourself. If you have any feedback (positive or negative) about your experience using OpenDC -during this tutorial, please let us know! |
