Welcome to the IBM Systems Workload Estimator Measured ...wle.mybluemix.net/wle/HTML/en/Tutorial/PMWalkthru.pdfWelcome to the IBM Systems Workload Estimator Measured Workload Integration
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Measured Workload IntegrationThe IBM Systems Workload Estimator provides sizing recommendations for an IBM system running one or more workloads associated with e-business and/or collaboration.
Customers using the Estimator should review all results with their IBM or IBM Business Partner representative.
This tutorial will introduce the reader to the flow of screens to expect in the Workload Estimator when coming from a measured workload source such as Performance Management for System i and System p or PTX. It is not intended to teach the reader how to do performance sizing. Only the basic input and navigation is included. For more detailed description of Workload Estimator use, features, and customization, see the "Detailed Walkthru and Features" tutorial.
Other Workload Estimator tutorials that are available include:"General Product Walkthrough““Detailed Walkthrough and Features”"Estimation Options""Consolidating Existing Systems“"Using Save and Re-use""Logical Partitioning (LPAR)““Obtaining and Configuring the AIX Provider (From PTX/Topas)”“Sizing Multiple Intervals”
• PM Workload• If you have the data in xml format for the workload, you may continue with this
walkthrough.• If you do not have data for the workload, please refer to Chapter 8 of this Redbook.
• PTX Workload• If you have the data in xml format for the workload, you may continue with this
walkthrough. However, keep in mind that the screens shown in the rest of this walkthrough have been taken while restoring a PM Workload so the screens you will see while restoring a PTX Workload will be similar, but not identical.
• If you are unsure if you have any data or not, check the /etc/perf/ folder and look for any files of the type [hostname]_aixwle_weekly.xml. Keep in mind that the Workload Estimator typically requires at least one week of data for a minimal estimation.
• If you do not have any data, please refer to the “Obtaining and Configuring the AIX Provider (From PTX/Topas)” walkthrough to find out how to get it yourself.
The remaining procedures list any differences between operation with PM eServer for System i/System p and the AIX Provider for System p.
From within PM, invoke the Workload Estimator by clicking on this “Size Next Upgrade” button. You will be transported into the Estimator with your PM Workload data already entered.
Alternatively, for systems not registered with PM, measured workload data can be imported directly into the estimator. This import method is required when using the AIX Provider for System p.
This is the name of the company transmitted by PM. If a company name is not available, "Not Available" will be inserted here. If you are using data from the AIX Provider for System p, the field will be blank. This field has no effect on the estimation.
The amount of memory that was installed when this data was collected. Some providers of information to the Estimator can supply actual memory usage data. If this is the case, the Estimator will also show this amount.
The most recent disk information is shown. Starting with v2006.1, the data providers can supply the Estimator with disk information in groups. These groups are typically things like “logical volumes”, or “ASPs”.
Each group is shown separately and will be treated as separate entities.
The amount of information that was provided to the Estimator. Also shown is the date of the most recent entry.
The data from the time period that is represented in the data is used by the Estimator to calculate the growth rates of the individual components of the performance data.
This section deals with analyzing the PM data that has been collected and sent to IBM or available in the AIX Provider for System p report. From this data the growth rates of the major system resources are calculated.
Initially this data reflects the growth trends that were calculated from the data itself. However, if you manually change any of these values, these will be used instead.
IBM System i:The "Interactive CPU" and "Non-Interactive CPU" fields are the growth rates for each of these two fields. These two fields are combined to arrive at the "Total System CPU" growth rate.
Note that the "Total System CPU" cannot be directly changed since it is a combination of the two components.
IBM System p:“Total System CPU" is the only CPU field and is editable.
Both:Remember that these values were initially calculated from the data itself. You can change them if you desire.
This is the growth in memory. Its growth rate is usually determined by the radio buttons on the right.
If one of the buttons after one of the other resources is checked, memory will grow in the same "percentage of the total“ as the resource that is selected.
Alternatively, you can select the "Grow Independently" in which case you will need to specify the growth rate. If the data actually contained Memory usage data, this value will be calculated from the data.
The growth information for each disk group is shown. Starting with v2006.1, the data providers can supply the Estimator with disk information in groups. These groups are typically things like “logical volumes”, or “ASPs”.
The “x-axis” is shown by date. Each date will usually show a green or black selection box underneath it. Green boxes mean that the data point is active and is to be considered the “least-squares fit”. A black box means that the data point is inactive and will not be considered in the calculation. The absence of a box means that no data for that point was supplied by the data provider.
A data point can be selected/deselected by clicking on the black/green box.
The resource to display in the graph by using these buttons. Click on one. It will turn green signifying it is the active one, and the graph is updated.
These are the current growth rates of the various resources. Initially most of the fields presented were copied from the previous screen of the Measure Data workload. If changes are made to them, they will be retained and copied back.
This column is the amount of each resource that is being "consumed" by the system to perform the work. This is calculated by looking at the latest month of data that the Estimator has access to.
The "Current Trend" is the current growth rate for the resource. It is calculated by "drawing a straight line“ (i.e. performing a least-squares fit) through the data. If data points have been deselected, that point on the line is ignored for the calculation.
The blue line on the graph is the representation of this line.
If you do not like the calculated "Current Trend", you can enter your own in here. Notice that as you enter in new growth rates, the dashed portion of the blue line changes to reflect the new "slope“.
If you have made any changes to the growth rates on the previous screen, those changes will be reflected here.
Normally the "Current Trend" and the "Future Growth" are identical. However, if you have changed the "Future Growth" from its calculated value, this connection is broken, and the associated arrow box will become empty (as shown in the top box).
Clicking on the empty white box will restore the connection between the "Current Trend" and the "Future Growth".
This column of radio buttons acts in the same manner as on the previous screen. Select the button next to the resource that you desire memory to "mimic". Memory will then grow at the same percentage of the totals as the selected resource is.
Alternatively, you can select the "Grows Independently" in which case you will need to specify the growth rate.
The handling of the Disk Arm resource is one of the trickiest resources to deal with in a general way. You will note that the graph is divided into 3 zones. The yellow "Over Utilized zone", the greenish "Potential Performance Degradation Zone", and the gray zone on the bottom.
As a general rule, disk arms should not be utilized more than 30% for optimal performance. The "Potential Performance Degradation Zone" is where the Estimator does not know if the utilization is low because you needed Arms to get the disk capacity, or if it is truly needed by the application.
Dragging the line upwards, you can see the percent utilized change in the yellow box on the left hand side of the graph.
What is actually happening is that you are allowing the disk arms to become more busy. If you feel that the work that this system is performing is NOT being negatively impacted by the number of disk operations occurring, moving this line upwards would be recommended when the starting utilization is significantly below 30%.
The Immediate Solution was chosen by taking the "upgrade path" information into account. You may wish to review the "Detailed Walkthrough and Features" tutorial for information about the various fields. As previously explained, this page also allows you to project your system growth and size into an appropriate system which will handle your business in the future. This 'future time horizon' for projection can be set in the "Advanced Growth Options" screen of the PM or AIX Provider Workload in the "months to grow" field. The default projection is 12 months.
If you have not had a chance to look at the other tutorials, they are highly recommended.
"General Product Walkthrough""Detailed Walkthrough and Features""Estimation Options"" Consolidating Existing Systems“"Using Save and Re-use""Logical Partitioning (LPAR)" “Obtaining and Configuring the AIX Provider (From PTX/Topas)”“Sizing Multiple Intervals”