janeteblake

My report from the technical conference.

For Day 3 session 3 I went to Better together: How Microsoft Dynamics AX leverages SQL Azure and the latest Microsoft data platform with Sunil Agarwal, Michael Gall, and Milinda Vitharana.

Lots of visuals today.

Agenda

• Topology in the cloud or on premise using SQL Server 2016
• Dynamic scaling of server in the cloud
• Multiple read-only databases to distribute the load
• One database for OLTP and Analytics workloads
• In-memory OLTP for highly performant operations
• Dynamics AX Entity store and scenarios
• Cortana Analytics suite integration

You can dial up or down the database to meet the load demand

No more cubes. Analytics are served by the real time database

Topology in cloud or on-premise

Dynamics AX platform

We’ll be talking about the data layer in the photo

SQL Server in Dynamics AX topologies

Back end is not SQL Server; it’s SQL Azure

Features released to the cloud before they are released to the box. Then customers test them then we send a high quality product to the box

Dynamic Scaling

SQL Database service tiers

Premium = Enterprise

Predictable performance

Microsoft scales up and down as needed; it is invisible to the user

Elastic database pools

Multiple read-only databases

Dynamics AX data

Primary to secondary in seconds

Secondary to entity store in minutes

Read only secondary (ROS) DB

• Available in production environments
  • One or more ROS DBs provisioned based on volume and load
• Selected workloads routed by AOS to ROS
  • Workload patterns determine whether they can be run on ROS. Developer doesn’t choose
• Some (limited) configuration options
  • Ex: Tile cache can be optimized by administrator where required
• Exact copy, seconds late…
  • Data in primary DB is reflected in secondary within seconds

Multiple Read-only Secondary Databases

Real-time operational

Minimizing data latency for analytics

Add challenge: Delivering Performant Analytics

Real-time analytics

NCCI is very efficient

Real-time analytics: Minimizing columnstore overhead

Demo: How AX leverages SQL Operational analytics

Index type can be index or columnstore

Future: Secure data on the move and at rest

Always encrypted

“like” is not available because similar strings encrypt differently

Temporal option available so you can restore from 5 or from 20 minutes ago

Future: in-memory OLTP

Hardware trends

In-memory OLTP – architectural pillars

Transform high level operations to C programs

30x performance improvement

SQL In-memory engine

How AX7 uses in-memory OLTP

In 2016 we allow 2TB for in memory

In the future: Dynamics AX entity store

• Mark the table as Storage Mode = InMemory
• Set Durability:
  • SchemaAndData: Data can be recreated after crash (from transaction log)
  • Schema: Data is lost after crash
• System administrator has to accept the setup

Dynamics AX data

Entity store – key scenarios

• High volume, near real-time Power BI reporting
  • De-normalized schema, CCI to enable faster query responses, incremental update from AX, direct Query Power BI models
• Intelligent business processes with Cortana Analytics suite
  • Stage data for Azure Data Factory pipelines, Consume in Azure Machine Learning
• IoT, external data integration
  • Staging in Azure Data Lake and Azure Data warehouse

Entity store – key tenants

• Designed for analytical scenarios
• Azure DB entity store included in AX license
• AX manages deployment and refresh
• Optionally, customer can stage data in Azure DB, DL and DW for read-write scenarios
• Coming soon to AX2012 R3

There will be an entity store for AX 2012

Future: Cortana Analytics Suite Integration

From data to decisions and actions

Cortana Analytics Suite

From data to decisions and actions

Updated

Example: Integrating AX Retail with Azure ML Recommender

AX + CAS go-to market approaches

CAS = Cortana Analytics Services

Summary

Happy DAXing!

My report from the technical conference.

And many, many thanks to those who retweet these blogs! It’s much appreciated. I am now pushing to complete this… look for this to be wrapped up by Monday morning.

I went to Mining the metadata and authoring X++ best practice rules in Microsoft Dynamics AX with Peter Villadsen, program manager.

It was full (of course) but I refused to miss another Villadsen breakout session, so sat on the floor. I started a trend and had lots of company.

Session Objectives and Takeaways

• Session Objective(s)
  • Learn about the role of metadata in Dynamics AX7 and how to leverage its value
  • Understand the extensibility story for the new X++ compiler
  • Understand how to write add ins into Visual Studio to provide new functionality
• The metadata is more pervasive and more useful than ever
• The AX development environment can be extended
• It is possible to extend the compiler easily

Dynamics AX Metadata

• The Dynamics AX environment is based on metadata
• Metadata is now stored in XML files in the file system, not in a database.
• Everything is open; there are no “blobs”.
• The XML files are the unit of Source Control.
• The metadata access does not depend on server

XML Metadata files

Showed them on the file system

Application explorer is a read-only view

Metadata access

Consuming metadata

• There are several ways of consuming metadata in Dynamics AX7:
  • Using raw System XML APIs
  • Using the dict* classes from X++
  • Using the automation API from Visual Studio add-ins
  • Using managed APIs: (compile time, general)
  • Using the OData service

Consuming metadata as XML

• You saw the XML metadata stored in files by package, model, and kind
• It is easy to do sweeps of these files using XML APIs in either X++ or C#.
• Pros:
  • XML can be accessed on any device, everywhere
  • Easy to search using LINQ or XPATH
• Cons:
  • It is not type safe
  • It is slightly clunky

Demo

Application note: large scale query map of metadata files. Used an open source edition, ran even faster

Using the dict* classes

• The dict* classes from AX6 were re-implemented on top of the new metadata API
• They support existing code, but new metadata is now leveraged through these APIs
• The TreeNode API is available too for some cases (except for writing)
• Pros:
  • Compatibility
  • Easily used from X++
• Cons:
  • A little kludgy
  • Inelegant but they do work

Consuming metadata through dict* classes

Demo

Legacy code will probably work

Accessing Data Entity metadata through ODATA

• It is possible to access Data entity metadata through services
• We can open this service up for other metadata artifacts as well, as needed
• Currently only open to entities. In the future will be tables

Using the managed metadata classes

• The managed metadata API are used throughout the platform
• They are classes that expose all the metadata in a type-safe, DOM-like way
• Pros:
  • The metadata can be consumed using state-of-the-art C# constructs
• Cons
  • There is a certain learning curve to negotiate
  • There are several different APIs, for different scenarios

Managed metadata access

• Metadata is accessed through a metadata provider
• There are three distinct kinds of managed metadata access:
  • From Visual Studio Extensions
  • From compiler scenarios
  • General metadata access APIs
• Traverse with LINQ or using custom code
• Stay tuned for examples…

Checkpoint

• You understand the concept of metadata and how it relates to Dynamics AX
• You have seen several ways to query the metadata
• Let’s do something useful with this knowledge!

Extending the Visual Studio experience through add-ins

Demo

Visual Studio

File > New > Project

DAX: BP tools, dev tools add-in

You get a template, just complete the TODOs

System.console.writeline(“stuff”)

iForm, iTable are the VS views

Metadata access

• You have the choice between two different metadata providers:
  • One is for accessing the MD as it is loaded in VS (using classes in the Microsoft.Dynamics.Framework.Tools.MetaModel.Automation namespace)
  • Example: ITable
  • The other is the global metadata store (using classes in the Microsoft.Dynamics.AX.Metadata.MetaModel namespace)
  • Example: AxTable

Compiler Extensibility

Simple parsing and analysis demo

Compile xpp – collect error messages

Create metadata provider

Get tabl emetadata

Compile class declaratino

Dump diagnostic messages

Run pipeline

Dump messages

“Macros are evil”

API is not elegant

Checkpoint

• You have seen how the metadata can be queried
• You now understand the concepts of Abstract Syntax Trees and how they relate to X++ source code.
• Let’s put this to good use!

Extending the compiler with BP rules demo

Visual Studio > File > New project > Use BP add in

Creates a template

There are white papers on writing BP checks

When you need to debug BP

Attach to XPPc agent

Check the wiki

Using the cross reference

• The cross reference database (DYNAMICSXREFDB) lives in the SQL database on your development box
• The format has been significantly simplified: it is mainly two tables
  • Names
  • References
• The cross reference powers the VS experience

Building the xref is relatively expensive but not absolutely expensive (4 minutes)

Q&A

Happy DAXing!

My report from the technical conference.

For the first session of the last day, I went to Ask the Experts: Developer and customization experience with Peter Villadsen, Robert Badawy,and  Joris de Gruyter. Quite the lineup!

Being an Ask the Experts, it had no set structure other than a Question & Answer format. Here are my scribbled notes on them. Note that it seemed everyone was sick & there was a lot of coughing during the questions and answers.

1. Replicated DB’s: adding indexes, etc.
2. Upgrade: prepare your code in 2012, split it in AX 7. Wiki has hints on code upgrade
3. One solution with two projects: cut and paste across projects
4. No need for remote based app / remote desktop
5. Security and data
6. DB replicas are read only, near real time. Can we query this? Not sure
7. SSRS still used, pretty unchanged. We don’t yet support extensions on reports
8. Jobs – now class – static void main(Args) and set as startup. Alternatively for data update you can create a data project. Use the data management framework
9. Alternative to SSIS: Whole new world of BI. Use Wiki > Analytics
10. PowerBI?
11. Move elements to different model? Yes. In projects via cut and paste. Alternatively just move the files. While you’re doing that, consider creating extensions.
12. Wiki > Developer > Code migration > Prepare to migrate 
13. Something how long it takes to run. Load balancing, zero downtime, deployment – automated on our end but manual work on Microsoft’s end. They are improving.
14. Printers – still support ZPL / zebra. Install an app locally that has access to printers
15. Extensions – there is a metadata API. It’s all file based.
16. How to develop. We do not recommend or support development in a shared AOS. Multiple developers can be on the same box. You can also duplicate the modelstore: Copy the packages folder in C:\ or J:\
17. XLNT framework? Use Raw API and somewhat raw API
18. Deploying deltas. If you overlay, it contains models underneath it
19. Unit tests. You can write unit tests. Remember that testing is very data bound. Microsoft test code is large
20. Multiple developers on the same box. Production database copied to development box – talks are ongoing
21. How browser connects to server. Use SysOperation framework. Use cancel button
22. Models, XML files are design time concept. Labels have no IDs now. You should name your labels
23. Explanation of packages and extensions. In the file system you see packages when you create an extension. You create a new package.
24. Copy of production database that’s not being refreshed for sandbox – the features exist, are currently being worked on.
25. Code refactoring. There are not very good tools. Macros cause issues.

Here I need to give a shout out to the guy with the great shoes! Keep on stylin’!

Happy DAXing!

My report from the technical conference.

This was the session where I mentioned ahead of time that I had a heart-rending decision to make. Fortunately, it was made for me.

I went to The new X++ compiler and the Microsoft Dynamics AX development platform with Peter Villadsen and Robert Badawy but it was completely full.

So I then went to Deep dive into Microsoft Dynamics AX packages and extensions with Joris de Gruyter. That had also been threatening to fill up, but they moved him to a bigger room.

Choice made, by default. I mean, everyone needs and wants this session! Go look it up!

Customizations = overlayering

Extensions

Some customers have reimplemented

Agenda

• Model Store > Packages > Layers > Models
• Customizations vs Extensions
• Extension Scenarios
• Upgrading to Extensions
• Tips and Tricks
• Extensions into the Future

Model Store > Packages > Layers > Models

• Layers and models are a design-time concept only; they mean nothing at runtime
• A package is one independent, individual set of layers and models; we are now compiling directly to dll’s
• A package is a set of folders, consisting of XML files representing the objects; XMLs sit in a folder according to the package and model they’re in

• A package as a whole translates 1:1 with a unit of compilation, i.e. an assembly (DLL)
• Packages can have references to other packages, just like .NET assemblies can reference each other. But they cannot be circular
• One or more packages combine into a deployable package, i.e. a unit of deployment

Demo: Packages, files and folders

“Model view” breaks it out by model

J:\AOSService\PackagesLocalDirectory\<Package>\<Model>

Descriptor folder is like the model manifest

Customizations vs Extensions

• Customizing (“over-layering”)
  • Granularity of over-layering is down to the property level. So different customizations can change different properties on the same control. You’ll never want to over-layer a menu.
  • “Delta” files store changes only, with their original value
• Models technically also “over-layer” each other
  • No more “traditional” model conflicts, only over-layering
  • Model conflicts don’t require changing the model’s contents
  • Over-layering granularity avoids most conflicts

Demo: Customizing, overlayering concepts, delta files

Project can only be in one model

Right click on the object and select customize or create extension

The XML stores both “original” and “yours” – if an ISV sends a hotfix, you now have a three way compare

Search bar: CF: = conflicts. Right click and resolve code conflicts

If you customize thru overlayering, upgrades will be far easier

Customizations vs extensions

• Extensions are the new customizations (however going to pure extensions is near impossible)
• Extension enablers
  • Events, metadata extension features, extension methods
• Extension packages
  • Benefits on a code base short and long term
  • Package split impact on customizations
• Change your extension suffix

Q: Can you turn an extension off? A: No
Demo: Extensions, capabilities

Everything extends platform, then foundation, then application suite

We use a lot of attributes

Name is: form.extension

Can rename to form.NewExtension

You can open an element in a specific project, package, etc. and that affects what you see.

Hierarchy of models? Last one installed wins

Extension Scenarios

• Adding table fields, UI changes
  • Table extensions, form extensions, data entity extensions
• Pre / post and other event handlers
  • Table event handlers, UI event handlers, traditional pre/post and [hookable]
• Customization / extension hybrids
  • Customize based on extension artifacts, delegates
• Plugins
  • Make your code extensible
• “Package split” effects
  • Customize multiple packages
• If a method is private, you can’t do event handlers on it unless you make it hookable.
• Plugins are like the extension framework in 2012

Upgrading to Extensions

• New artifacts vs customized artifacts
  • Moving new artifacts has consequences
• Refactor vs upgrade
• Current limitations
  • Queries (you can’t really extend a query), over-layering extensions
• Security
  • Roles, duties – not privileges (you can’t extend privileges)

Tips and tricks

• Metadata search
• Go to form BaseForm
• Quick way to move a control on a form is cut and paste
• Tags
• Visual Studio has bookmarks
• Projects are not in your AOT; they’re just files in your file system
• Reports don’t allow extensions
• Element IDs are unnecessary now

Shout out to Yagesh! And sorry if I spelled your name wrong. Thanks for saying hi.

Happy DAXing!

Photo: Olympic Sculpture Park. Credit: me.

My report from the technical conference.

Sorry to say that the fourth session on the second day was a letdown.

I attempted to get into The X++ language with Peter Villadsen but it was full – completely, overflowingly, full. They need to get him into bigger rooms. I gather it is currently available at mbspartner.microsoft.com/AX/Topic/48  but I don’t have Dynamics Learning Portal so I’m out of luck for now.

So then I went to a lab on Applying form patterns with Jason Green. It had a lot of potential, but too many wrinkles to work out. I ended up leaving. But here was the overview of the lab:

Lab: take a 2012 form and apply a pattern

Lab overview –

• Apply the details master form pattern

• In Solution Explorer, open a form
• Right click, set as startup object
• Press Ctrl-F5 to build/run
• Open the Patterns Information Panel by clicking the Patterns tab in the form designer
• Right click Design, point to Apply pattern, then select the pattern (details master)
• Add missing groups and controls
• When done, press Ctrl-F5 to build/run

• Utilize documentation to learn about patterns

• Help is at the AX Help Wiki at http://ax.help.dynamics.com; search for “details master” and open the “Details Master – Form Pattern” article

• Apply subpatterns to form controls
  • In VS, in the form designer, right click the form, select Addins, and select Form statistics
  • You will see coverage – look for Pattern=Unspecified to be zero
  • Use the VS form search to find all instances of “unspecified” in the form
  • Subpatterns are part of the Selecting a Form Pattern wiki page
  • Right click Comments, select Apply pattern, and select your pattern
  • Repeat until there are no more unspecified
  • Press Ctrl-F5 to run
• Use patterns-related Visual Studio add-ins

• In VS, click Dynamics AX, Addins, Run form patterns report
• You’ll get a notification of where the csv went; open it in Excel
• Filter it to your model and/or form

Shout-out tonight to Bobby Small – thanks for reading my blog and taking the time to say hello!  Bobby is @daxioms on twitter and daxioms.blogspot.com.

Happy DAXing!

Photo: Washington Park Arboretum Credit: me

My report from the technical conference.

Our third session on the second day, Wednesday February 24, 2016, was a breakout session, called  Microsoft Dynamics AX implementation methodology, with Clay Wesener and Paul Wu, of the SA group.

I hope you can make sufficient sense of my scribbled notes. This was definitely a lot to take in but of course very important. We really need to be comfortable with this. You may have noticed, I’m trying not to infringe on Microsoft’s slide decks, but I think they’re necessary for clarity in this case.

Implementation methodology

There is now a customer LCS project

This is about the tools

Implementation using CRP

• Smaller iterative approach
• Easier to spin up
• Also see implementation lifecycle workshop

LCS based tooling

Discover-define-develop-operate

Business process management is much improved – it can support off-AX things

VSTS – business processes, traceability requirements, change requests, deliverables (FDD/TDD), builds, dev work, bugs, test cases, support tickets

LCS – BPM, usage profiles, upgrade assessments, subscription estimate, deployable packages, user task guides, process data packages, configuration manager, cloud powered support, issue search, monitoring and diagnostics

If you get nothing else from this post, then “get” this chart:

Demo: Deploying an environment (Paul Wu)

Demo: BPM and task guides (Clay)

LCS > Business process libraries

You can edit a recording

You get an APQC library with AX

Usage profiles and subscription estimator

LCS > Subscription estimator

Can monitor a VM, but can’t dynamically adjust it

BPM, process data packages (associates DP with a process in your BPM library), configuration Manager

BPM library

You actually load the data into AX from LCS

LCS > Data management

LCS > asset library: you can put in data packages

DIXF – configuration manager – process data package

We can specify dependencies (sequences). They advise, not enforce.

TR code gets checked into VSTS

Support and operations demo (Clay)

Happy DAXing!