From 0f835d57b0e989e25aa0b71fe374a0fb1a94e86f Mon Sep 17 00:00:00 2001 From: Dante Niewenhuis Date: Tue, 5 Nov 2024 14:17:08 +0100 Subject: Documentation update (#261) * Updated a lot of documentation, added a new get-started tutorial. * Applied Spotless * Applied Spotless Java * Added bitbrains workload to site --- site/old_tutorials/2-experiment.mdx | 74 +++++++++++++++++++++++++++++++++++++ 1 file changed, 74 insertions(+) create mode 100644 site/old_tutorials/2-experiment.mdx (limited to 'site/old_tutorials/2-experiment.mdx') diff --git a/site/old_tutorials/2-experiment.mdx b/site/old_tutorials/2-experiment.mdx new file mode 100644 index 00000000..14970ea6 --- /dev/null +++ b/site/old_tutorials/2-experiment.mdx @@ -0,0 +1,74 @@ +--- +description: How to experiment with your datacenter designs. +--- + +# Create an Experiment + +After designing a datacenter in OpenDC, the next step is to experiment with the design using OpenDC's built-in +simulator, in order to analyze its performance and compare it against other designs. + +In OpenDC, we use the concept of portfolios of scenarios to experiment with datacenter designs. In the next sections, we +will explain how you can use these powerful concepts for the experimental analysis of your designs. + +## Create a Portfolio +OpenDC organizes multiple scenarios (experiments) into a **portfolio**. Each portfolio is composed of a base scenario, +a set of candidate scenarios given by the user and a set of targets (e.g., metrics) used to compare scenarios. + +To create a new portfolio, open a project in the OpenDC web interface and click on ‘+ New Portfolio’ in the top right +corner of the portfolio table. This opens a modal with the following options: + +1. **Name**: the name of your portfolio. +2. **Metrics**: the metrics that you want to use to compare the scenarios in the portfolio. +3. **Repeats per Scenario**: the number of times each scenario should be simulated (to account for variance). + +## Create Scenarios + +A **scenario** represents a point in the datacenter design space that should be explored. It consists of a combination +of workload, topology, and a set of operational phenomena. Phenomena can include correlated failures, performance +variability, security breaches, etc., allowing the scenarios to more accurately capture the real-world operations. + +The baseline for comparison in a portfolio is the **base scenario**. It represents the status quo of the infrastructure +or, when planning infrastructure from scratch, it consists of very simple base workloads and topologies. +The other scenarios in a portfolio, called the **candidate scenarios**, represent changes to the configuration +that might be of interest to the datacenter designer. Dividing scenarios into these two categories ensures that any +comparative insights provided by OpenDC are meaningful within the simHyperVisorContext of the current architecture. + +To create a new scenario, open a portfolio in the OpenDC web interface and click on ‘+ New Scenario’ in the top right +corner of the scenario table. This opens a modal with the following options (as shown in [Figure 1](#explore)): + +1. **Name**: the name of your scenario. The first scenario of a portfolio is always called the _Base scenario_. +2. **Workload**: the applications (e.g., virtual machines, workflows, functions) that consume resources in your + datacenter. + 1. **Workload Trace**: A dataset that characterizes the historical runtime behavior of virtual machines in the + workload over time. + 2. **Load Sampling Fraction**: The percentage of the workload that should be simulated (e.g., 10% of virtual + machines in the workload trace). +3. **Environment**: + 1. **Topology**: one of the topologies of the project to use for the scenario. + 2. **Scheduler**: the algorithm that decides on which hosts the virtual machines should be placed. +4. **Operational Phenomena**: + 1. **Failures**: a flag to enable stochastic host failures during simulation. + 2. **Performance interference**: a flag to enable performance interference between virtual machines (only available + for a subset of traces). + +Once you have created the scenario, it will be enqueued for simulation. Usually the results of the simulation should be +available within one minute after creation. However, if there are lots of queued simulation jobs, it might take a bit +longer. + +
+ Creating a new scenario in OpenDC +
Creating a new scenario in OpenDC. The user can select the topology, workload, and operational phenomena.
+
+ +## Analyze Results + +After creating scenarios, the scenario table will show whether the simulation is still pending, completed successfully, +or failed for some reason. If the scenario was simulated successfully, its results will become visible on the ‘Results’ +tab as shown in [Figure 2](#analysis). + +
+ Analysis of results reported by OpenDC +
Plots and visual summaries generated by OpenDC comparing different scenarios.
+
+ +This tab will show the selected metrics for the portfolio and allow you to compare their values for different scenarios. -- cgit v1.2.3