Performance – Devlicious.net

Demand Forecasting: Performance influencers

In this blog post we will explain the key performance influences for the Demand Forecasting feature for Microsoft Dynamics 365 for Finance & Operations, which are under the user’s control.

It is important to read and understand this blog if you are implementing Demand Forecasting for Microsoft Dynamics 365 for Finance & Operations or if you are experiencing the a timeout during generation of the statistical baseline. The error message will be:

Forecast generation in Azure Machine Learning timed out

Number of enabled forecasting dimensions

The Forecasting dimensions are accessible via Demand forecasting parameters form under the Master planning menu, as illustrated below:

Enable only the must have forecasting dimensions. If you experience a timeout in the Azure Machine Learning when generating the statistical baseline forecast, one of the first things to consider is, if any of the enabled forecasting dimensions can be disabled.

Number of items in Item allocation keys

The Item allocation key form is accessible under the Master planning menu, as illustrated below:

Each allocation key will currently be executed in isolation on the Azure Machine Learning web service, this implies that the number of items in each allocation key, will have a direct impact on the performance of processing each allocation key.

Typically, which items are assigned to which item allocation key is a business decision, however to the extent possible try to keep number of items in each group below 1 thousand or at least in the low thousands.

Historical Horizon

The historical horizon is assigned before running the generation of the statistical baseline forecast.

The size of the historical horizon has a direct impact on the performance of the generation of the statistical forecast.

It is worth considering if historical data is representative of what is expected in the future. For example, 3 years of historical data may be available, but in some cases it can be better (result in a more accurate forecast) to use just last year’s historical data. It requires insight into the historical data to make this decision but is something that should be considered.

Selecting a single forecast model

As a last resort a single forecast model can be selected, the options are:

STL
ARIMA
ETS

Product Configuration: Performance with Z3

Introduction

In the spring release of 2018 for Microsoft Dynamics 365 for Finance & Operations a significant investment in improving performance of the constraint evaluation has been made.

The result of this investment is the introduction of the Z3 sovler strategy. Z3 is a theorem prover from Microsoft Research, that has won several awards and has an active open source community.

The Microsoft Dynamics 365 for Finance & Operations team, teamed up with the Microsoft researches to develop new algorithms targeted at the Product configuration scenarios for Microsoft Dynamics 365 for Finance & Operations.

Microsoft Dynamics AX 2012 R3

The Z3 solver strategy is also available for Microsoft Dynamics AX 2012 R3 with the following KB article numbers:

4341629 - Additional performance improvement of loading models

4046549 - Performance improvement of loading models

4034024 - Fix to support customization related to system defined table constraints

4012892 - Introduction of the Z3 Solver strategy

Usage

The applied solver strategy for a product configuration model is selected by accessing the Model properties form.

From the Model properties form the Z3 solver strategy can be selected as illustrated below:

The Z3 solver strategy will currently not have a functional impact, the impact will be on performance.

Microsoft Dynamics AX 2012 Product Configuration : Enterprise Portal

In this blog post I will explain how on user interactions are handled by the product configuration module for Microsoft Dynamics AX 2012 on the Enterprise portal.

The figure above, is a high level illustration of how user interactions in general are processed on the Enterprise Portal (EP) for Microsoft Dynamics AX 2012. The clients access the AOS through EP, which is a web implementation built on top of Microsoft Sharepoint, hosted on a web server (IIS). The communication between EP and the AOS is synchronous using the .NET Business Connector RPC call. Each RPC call must be completed and returned, before http request timeout is exceed on the IIS.

To understand what implications the above has, when using the Product configuration module, we will go through the sequence of events, which take place when the user loads a configuration model and assigns a value to an attribute.

The above illustrates the sequence of loading a configuration model and assigning an attribute a value on the Enterprise portal.

Loading a configuration model

When the user wants to configure a product, e.g. from a sales order, the requests will be handled by EP. The EPPCConfigurator web control will be loaded, inside the web control a proxy Instance of the PCEnterprisePortalMain class is created. This proxy is used to interact with the AOS Product configuration logic.

On the AOS, the XML representation for the specific configuration model is retrieved from the database. Then the XML string is passed as argument to Product Configuration Server API, which unlike the Client API is design to support synchronous execution. In the creation of the ProductConfigurationServer instance, the XML is parsed and supporting data structures are created, just like on the Windows client implementation.

Once the data structures have been created, the configurator instance can start processing the constraints and calculations of the configuration model. Unlike the Windows client however, the processing on EP needs to respect a fixed time limit, in order not to cause a timeout on the web server.

Another difference between the Windows client an EP is the way the model is being processed. The EP implementation will attempt populate all visible enumeration and Boolean attributes. This is the equivalent of opening all the visible attribute controls on the active component in the form, and letting the configurator deduce feasibility of the values of each attribute.

The more attributes that are visible, the less time that is available to process each attribute. For components with over 20 enumeration and boolean attributes, it can happen that there is not enough time to process the feasibility of all values within the domain of the attribute.

Assigning values to attributes

When the user makes an assignment to an attribute, the web form is then posted back to EP and a request to assign the attribute is made on to the AOS. On the AOS, each call from EP executes in its own context, this implies that a new instance of the configurator is created. This instance must again parse the XML representation of the configuration model, before being able to process the user assignment. After this, the instance is disposed and a new one is created to process the assignment , same as before.

The main problem can be seen from the diagram in Figure 2, is the activation line of the ProductConfigurationServer object. Unlike the Windows client implementation , the configurator instance is not kept alive throughout the configuration session. The synchronous execution combined with the time limitation is the main limiting factor, which forces us to create and dispose instances of the configurator for each request.

The consequence of this is that the EP implementation of the configuration module, is not able to deliver a user experience which is similar to that of the Windows client for models of medium to high complexity.

Please keep this in mind when considering implementations of the Product Configuration module on Enterprise Portal for Microsoft Dynamics AX 2012.

Checklist for improving MRP performance – Part 2: How to setup planning parameters

If Master Planning (MRP) runtime is too long, here are a couple of simple things you can try in order to improve it. You must, of course, try these out on a test environment first and, if you see an improvement, apply the same settings to the production environment. It is recommended that you apply one setting at a time, so you can get a good sense of what helped and what didn’t. Some of the settings below will not only have an effect on runtime, but may also improve your plan quality. The list was compiled for Microsoft Dynamics Ax 2012, so parts of the checklist may not apply to earlier versions.

1. Items assigned to the same coverage group should have similar lead times

Coverage parameters, such as the coverage time fence, negative days and positive days, relate to the lead times of the items inside the coverage group. If there is a disconnect between the lead times of the items and these parameters, master planning may create unnecessary planned orders, leading to potential suboptimal planning outcome and potential longer execution times. Items assigned to the same coverage group should have similar lead times.

Be sure at all times that you have the latest hot fixes related to lead times applied to your environment. One example is KB article 2938941: https://mbs2.microsoft.com/Knowledgebase/kbdisplay.aspx?WTNTZSMNWUKNTMMYZWLSWZQTLLRPPXLZNRORZMMMSSMTMNPU

2. Master planning parameters – Use dynamic negative days parameter should be checked

The negative days setting is connected to the lead time of the items. If the negative days are less than or equal to the item’s lead time, unnecessary planned orders may be created. Consider the following example: current date is 1st May, there is a sales order for 1 piece of item A on 5th May, item A has a lead time of 6 days and there is a purchase order for 1 piece of item A on 8th May. The negative day setting is 2, which is less than the item A’s lead time. Master planning will create a new planned purchase order for 1 piece of item A, which, according to the lead time will be delivered 7th May. This is just 1 day before the already existing purchase order. Firming this new planned purchase order may lead to unused inventory piling up. Creating planned orders that are not useful increases master planning runtime for no reason.

Checking Use dynamic negative days parameter makes sure that the connection between the item’s lead time and the negative days is taken into account during planning. Consider the following example: current date is 1st May, there is a sales order for 1 piece of item A on 5th May, item A has a lead time of 6 days and there is a purchase order for 1 piece of item A on 8th May. The negative day setting is 2, which is less than the item A’s lead time. Master planning will not create any planned purchase orders. The sales order will be fulfilled using the existing purchase order.

3. Negative days parameter on the coverage group should not be less than the lead time of the items assigned to the coverage group

If the negative days are less than the item’s lead time, unnecessary planned orders may be created. Consider the following example: current date is 1st May, there is a sales order for 1 piece of item A on 5th May, item A has a lead time of 6 days and there is a purchase order for 1 piece of item A on 8th May. The negative day setting is 2, which is less than the item A’s lead time. Master planning will create a new planned purchase order for 1 piece of item A, which, according to the lead time will be delivered 7th May. This is just 1 day before the already existing purchase order. Firming this new planned purchase order may lead to unused inventory piling up. Creating planned orders that are not useful increases master planning runtime for no reason.

Negative days parameter on the coverage group should not be less than the lead time of the items assigned to the coverage group.

A KB article you should make sure you have installed is 2736992 https://mbs2.microsoft.com/Knowledgebase/kbdisplay.aspx?WTNTZSMNWUKNTMMYMWMSYTMOWZPWLSWNTQWWSZZVXMWPMSOX

4. Positive days parameter on the coverage group should not be 0 or empty

If the positive days parameter on the coverage group should is 0 or empty, then existing incoming orders will not be taken into account for planning and new planned orders will be created. Note that on hand will always be consumed, regardless of the number of positive days set. Creating planned orders that are not useful increases master planning runtime for no reason. Firming these planned orders may lead to unused inventory piling up. Positive days parameter on the coverage group should not be 0 or empty.

5. Items that are regularly procured or produced should be assigned to coverage groups where the Positive days parameters is equal to the item’s lead time

If items have lead times bigger than the Positive days parameter on the coverage group, unnecessary planned orders may be created. Creating planned orders that are not useful increases master planning runtime for no reason. Firming these planned orders may lead to unused inventory piling up. Items that are regularly procured or produced should be assigned to coverage Groups where the Positive days parameters is equal to the item’s lead time.

Checklist for improving MRP performance – Part 1: How to run MRP

If Master Planning (MRP) runtime is too long, here are a couple of simple things you can try in order to improve it. The list was compiled for Microsoft Dynamics Ax 2012, so parts of the checklist may not apply to earlier versions. You must, of course, try these out on a test environment first and, if you see an improvement, apply the same settings to the production environment. It is recommended that you apply one setting at a time, so you can get a good sense of what helped and what didn’t. It is also recommended that the test environment is as close as possible to the production environment in terms of data volumes: number of products and number of transactions. If you are running MRP in recurrent batch, note that some of the settings will be cached, so it’s better to make the parameter change and re-create the batch and then run MRP again.

1. Use helpers for master planning

Using helpers (parallel processing threads) during master planning will most likely improve the runtime. Request one or more helper threads when running master planning, by setting the Scheduling helpers – Number of helpers parameter in the Master scheduling dialog to a value greater than zero.

More information about helpers can be found here:

TechNet: http://technet.microsoft.com/en-us/library/gg242496.aspx

InformationSource: https://informationsource.dynamics.com//RFPServicesOnline/Rfpservicesonline.aspx?DocName=TC2014+Help!++MRP+is+slow%7cQJ4JEM76642V-8-1608

Also make sure you have applied to the Ax environment the latest hot fixes related to helpers. Below are just a couple of KBs that you should consider installing or porting to your AX version:

KB article 2693422 https://mbs2.microsoft.com/Knowledgebase/kbdisplay.aspx?WTNTZSMNWUKNTMMYTPNRYVZNNPWYSWVQRWURNYWNXVKVRNQY

KB article 2690230 https://mbs2.microsoft.com/Knowledgebase/kbdisplay.aspx?WTNTZSMNWUKNTMMYTPNRYVZNNPWYSWVQUOLKVSMVQULNVYQT

KB article 2862912 https://mbs2.microsoft.com/Knowledgebase/kbdisplay.aspx?WTNTZSMNWUKNTMMYVWQXNRNXTVKNSKYVQNYNRZQRRRTPMQYY

2. Number of helpers used during master planning needs to be less than or equal to the maximum number of threads allowed on the batch server

It makes no sense to have the number of threads requested by planning greater than the sum of available batch threads on all batch servers in the selected batch group that runs master planning. Set the number of master planning helpers to be lower than the sum of available batch threads on all batch servers in the selected batch group to run master planning or increase the number of threads on selected batch servers.

To check the number of available batch threads per server/batch group go to System administration – System – Server configuration

3. Use multiple AOSs for master planning

Using more AOSs during master planning will most likely improve the planning runtime. It is also very important that you make sure no other lengthy routine is running simulataneosuly with master planning. It is also very important that you make sure no other routine requireing locks on tables used by master planning (for example InventTrans, InventSumLogTTS, ReqItemTable, etc) is running at the same time as MRP.

4. Configuration keys that are not actively used should be disabled

For example, when Process Industries functionality is used, a lot of additional planning logic is being triggered, like for instance expiration dates for batches. This will increase the master planning runtime. If Process Industries functionality is not used, make sure the ‘Process manufacturing’ configuration key is switched off.

5. Action message usage

Calculating action messages leads to longer master planning runtime. If action messages are not analyzed and applied on a regular basis (daily, weekly, etc), consider disabling the calculation during the master plan run, by making sure the Action message time fence is 0 on the master plan you are running (Master planning – Setup – Plans – Master plans). Also make sure all the coverage groups have Action message setting disabled.

6. Futures usage

Calculating futures leads to longer master planning runtime. If you are not planning BOMs or if propagating delays from supply to demand during planning is not needed for you, consider disabling futures calculation during MRP, by making sure the Futures time fence is 0 on the master plan you are running. Also make sure all the coverage groups have Futures setting disabled.

7. Number of tasks in a master planning helper task bundle

Changing the number of tasks in the task bundle may have a positive effect on the runtime. The number of tasks in a bundle controls how many items are planned together by a single helper. It is recommended to increase the number of tasks when the number of items is very large (hundreds of thousands) and decrease the number of tasks otherwise. In most cases it is a trial and error approach before an optimal value is determined. Remember to change this value as you add to Dynamics Ax more and more products!

8. Caching during master planning should be set to Maximum

Master planning logic is controlling the amount of caching used. This parameter is obsolete and is a candidate for removal in future versions of Microsoft Dynamics AX. In all the companies master planning is executed, consider setting Master planning parameters – Use of cache to Maximum