Visualizing your Workload Characteristics
The analysis dashboard for your Production Workload Data displays overall the workload I/O characteristics in the uploaded data, and it then identifies and characterizes the individual workloads contained in the data. The dashboard updates in real-time as the file is analyzed. You can stop the parsing of the file at any time by clicking on the red "Stop Parsing File" button in the top right hand corner. The status bar shows the analysis stage and the remaining time before the analysis is complete. Any errors will be logged in the log.
The analysis dashboard provides you a visual representation of your IOPs, Latency, Throughput and other workload characteristics.
The top section shows the status of the workload analysis and approximately how much time remains. It will show if there are any issues gathering the data and has a link to a Log in case issues arise.
The Access Pattern shows the IOPs, Latency (if available) and Throughput. You can zoom into any time segment on the chart by selecting a time segment on the chart with the mouse or view menu at the top of the access pattern section. The selected time frame will update all charts on the analysis dashboard. The convention used throughput the analysis dashboard is Blue for read operations, Green for write operations and Orange for all other operations.
For IOPs the Attempted line is a dashed line indicating that the work is not complete, the Successful line is solid indicating it has completed and an symbol is used to indicate any unsuccessful transactions in order to ensure they are visible (there shouldn't be any in a real production environment). If there are only a few failures you can determine whether they are with Reads, Writes, or Other commands by looking at the Errors summary card in the Summary View below or by turning of the other lines using the legend.
You can change switch the chart to a log scale or export the chart to pdf, csv or png from the wrench icon on the chart (top right).
The Summary View gives a good overview of the Production Workload Data and gives an overview of the type of workload that would be created if only a single Summary workload is selected. These values will automatically change if you select a specific time window from the IOPs, Latency or Throughput charts in the Access Pattern section.
The Summary cards give overall throughput, IOPs, Latency, and Errors broken down by Reads and Writes.
The data in the analysis screen is dependent on what is contained in the data you upload. If you see charts that say "No data available" this is a result of the data not being contained in the data uploaded.
- Read / Write Mix: display the read / write ratio observed in the workload data. This value is used to determine read/write ratio in created workloads.
- Request Size Distribution: shows the request sizes that have been seen and their average IOPs by size. It gives you insight into the block sizes configured in the servers and applications communicating with the storage infrastructure.
- Command Mix: gives the ratio of all of the commands that have been seen. When trouble shooting an issue it is very important to determine if there are any unexpected commands or if ratio is above what it should be. Knowing if certain commands are present may be all that is needed, as some of the commands can be quite disruptive and aren't normally seen in production environments.
- Average request size over time: gives a good indication of the changing relationship between IOPs and Throughput. Batch activities will typically have higher average request sizes as will certain transactions like video streaming or applications configured with large block sizes.
- Latency verse IOPs: provides an indication of whether the latency is limiting the IOPs or whether the IOPs are causing latency. It is a scatter plot with a trend line often the scattered dots will be appear as a line or two because of their relationship.
- If the line is trending up this is generally and indication that the latency is caused by the IOPs but isn't significantly limiting it. If the line is trending down this may indicate the latency is limiting the number of IPs that can be performed.
- Latency by block size: provides an indication of how much the request sizes are impacting performance.
- All things being equal a block size that is twice as big as another should take twice as long. You may see that is not the case and it can be normal.
- For example you may see that 4KB reads take the same amount of time as 8KB reads. This is generally an indication that the array is configured with an 8KB block size and is actually access all 8KB when only 4KB is requested. It might be a good idea to consider aligning the server block size settings with the arrays by changing one or the other.
Creating Summary Workloads from Production Workload Data
From a completed Production Workload Data, you can generate a Summary Workload from it by clicking on the Create Summary Workload button.
All Workloads created by WorkloadCentral are designed with to be used with Virtual Instruments (formerly Load DynamiX) Load Generation Appliances.
The following options are provided as shown below:
These options allow you to determine the amount of load to generate when running the modeled workload in a test environment. The option you select will be reflected in the Load Properties section of the Workload configuration on the next page. Regardless of which option you select, you still have the ability to customize any of the parameters that are exposed in the Workload configuration page (see the section "Creating a new Workload test" for details) before starting a test. The option you select simply modifies a set of defaults in the Workload configuration page so that you do not have to manually enter the information.
Analysis Result / Workload Components View
The Workloads Components View shows the discrete workloads that were identified by WorkloadCentral's workload analytics. For each workload identified the key characteristics are shown including how many Target/LUNs. The Quality % indicates the relative accuracy of workload identification that was determined by the analytics.
The workloads can be renamed inline to reflect better reflect the application or component they represent.
Creating Composite Workloads from Production Workload Data
From a completed Production Workload Data, you can generate a Composite Workload from it by clicking on the Create Workload button. A workload will be created for each of the components selected and named according to the label in the first column.
A Composite Workload will also be created that includes all of the selected workloads. Once the workloads have been created the user will be redirected to the newly created composite workload. Each of the component workloads should be mapped to a Test Bed connection with the number of LUNs clustered from the production workload.
Accessing and Viewing Your Workload Data
When you first upload the data it will take you straight into analysis dashboard for the Production Workload Data you just uploaded.
If you wish to view the results of your previous workload data uploads simply select Analyze Your Production Data on the Homepage and it will take you to a listing of your uploaded data and their status.