Tag: Azure (page 1 of 1)

My Least Favorite Thing About Azure Databases

Now that I have been working in the Azure cloud environment for a few months with my new job, I have come to the realization that there is one thing related to Azure SQL Server instances that annoys me more than anything else in the Azure environment. Obviously every cloud environment is going to have its pros and cons, so this is only one small problem from the whole ecosystem. Azure has a lot of nice things to work with, so this complaint is very superficial, but I know others at my company have been confused by this issue so thought I would share the annoying difference here for educational purposes.

On-Prem SQL Server Permissions

The one thing that has been driving me crazy about Azure SQL Server instances recently is that there is not an easy way to quickly get an overview of the permissions a user/account/group has on the server and database.

In on-prem/normal versions of SQL Server, you can open the “Security” section of the server or database then right-click on the user or group you want to review the permissions for, and then you’ll get a GUI that quickly shows all the server- or database-level permissions that user has. The following screenshot shows the Server Roles for a given login on my local test server, so you can quickly know what permissions the login has.

And then if you go to the “User Mapping” page of the Login properties, you can see any databases the login has access to as well as the permissions that user has on the database.

I love this interface that I’m pretty sure has been around for decades. It’s so much easier to view and quickly change the permissions for a login/database user that new new system in Azure-based servers.

Azure SQL Server Permissions

In comparison, you can’t easily view or determine the server- or database-level permissions any login or user has on an Azure-based SQL Server. You can’t even view any general properties of the user like you can for on-prem installations. For example, here is a screenshot of what you see on an Azure SQL Server when you right-click a Login from the Object Explorer:

If you do the same thing on the on-prem SQL Server, you can see that there’s an option to view the properties of the user, which will get you to the menus where you can see server- and database-level permissions, along with other properties as well.

If you can’t get a nice GUI representation of those login permissions, how are you supposed to know what permissions anyone has on your server or database? A SQL query.

While that kind of makes sense for a platform where you do most of your work with SQL, it does mean you need to do more work to write or track down a copy of the appropriate SQL query to get the necessary data, instead of to right-clicking on an object and have its information displayed to you immediately. The SQL query you need to get permissions information is also tedious to me because you can only do it for a single database at a time instead of being able to quickly view the permissions for every database on the server in one place. For example, if you have 3 databases on your server, you need to run the query individually for each of those databases. You can’t even utilize the “USE [database]” statement to quickly switch between the databases and get all the results in a single result pane (if you want to read more about that, I have another upcoming post to discuss that topic).

This is the query I use to review database permissions:

SELECT DISTINCT pr.principal_id, pr.[name], pr.[type_desc],
    pr.[authentication_type_desc], pe.[state_desc], pe.[permission_name]
FROM sys.database_principals AS pr
JOIN sys.database_permissions AS pe
    ON pe.grantee_principal_id = pr.principal_id
ORDER BY pr.name;

If you want to view the permissions at the server level, run that query on the master database. Then to view the permissions specific to a single database, you will need to switch to that database (using the database dropdown from the toolbar above the query window, or updating your full server connection to specify the other database).

To view the role assignments for users in your database, you can use the following query, which utilizes the views sys.database_permissions, sys.database_principals, and sys.database_role_members to gather all the useful information related to role assignments.

SELECT u.[name] AS [UserName],
       r.[name] AS RoleName 
FROM sys.database_principals u
     JOIN sys.database_role_members drm ON u.principal_id = drm.member_principal_id
     JOIN sys.database_principals r ON  drm.role_principal_id = r.principal_id
WHERE u.[name] = N'TestUser';

Summary

While it’s not hard to get all role assignments and assigned permissions for logins and database users in Azure SQL Server instances, it isn’t as easy as it used to be with on-prem servers. I don’t think the new method of getting the permissions gets the information you need in as useful a way as the old method of using the GUI. I personally don’t prefer having to scroll through a table of information about each user and deciphering the details of their permissions, but I will continue to deal with it since that’s the only option for Azure SQL Servers currently.

How to Download SQL Server 2022

You would think that figuring out how to download the installer for SQL Server 2022 Standard or Enterprise editions would be a simple task, much like the process for how you download any other installer or ISO for any other version of SQL Server. But with 2022, Microsoft seems to have changed their method for getting the software to people, making it a lot more locked down than other versions are. I’m not sure why this change was made, and if you have some insight please feel free to share it in a comment, but it was made and I have finally figured out how to get that installer with this new system. It took my colleagues and I a whole work day to figure this out. We had to work with our Microsoft account managers to get to the bottom of it, and it seems a little crazy to me that it would be that unintuitive to find the ISO for this software.

What’s in this post

Background

In my company’s current architecture, we create SQL Server virtual machines (VMs) by creating standard Windows servers through the Azure portal, and then one of the database developers will logon to the VM and install SQL Server through the ISO/installer, instead of creating a SQL Server VM from a provided image in Azure. In the past, when we needed to install SQL Server onto a VM, we would google something like “SQL Server 2019 download” and click on the link that takes you to the Microsoft downloads page, and then get the ISO from there. When we tried to do the same for SQL Server 2022, we were taken to a downloads page that only allowed you to download Developer or Express editions, and if we wanted to do Standard or Enterprise, we could only get an evaluation version after providing our information to start a free trial.

We obviously did not want a short trial version of the software, we wanted the full Standard edition, but could not find a full installer/ISO for that version. For a few months, we gave up and installed 2019 instead of 2022 whenever we needed to create a new server, but we reached a point where we are required to use 2022 for some applications, so I set out to find out how to get the Standard edition of SQL Server.

The Proper Method

There are two different ways that we found we could get a Standard or Enterprise edition of SQL Server 2022, and I’ll start with the method that I think they’re intending people to use. If your organization uses SQL Server in Azure already, you will likely need to go through this same method to get the installer for 2022.

I unfortunately can’t get any screenshots of this method of the process because I’m not an elevated user in our organization, but there is a way for your Azure tenant owner (and a few other select individuals) to get the latest ISOs and installers for every piece of Microsoft software your company has licenses for. I believe it is at admin.microsoft.com, but you can talk to your account representative to find out for sure. You can then navigate to “Billing” and then “Your Products” and then “Volume Licensing” where you can view all your licenses and get the downloads for them.

This is how we resolved the installer issue in our organization; we had our tenant owner download the ISO from this location and save the file in a shared location. Going forward, whenever we need an installer for any Microsoft product that we use, we will be required to have that one person download the files and send it to us (not a super convenient method if you’re a developer that doesn’t have full access to your entire Azure tenant).

Once you have the ISO, you can run it as you usually would to install any version of SQL Server, and your license key will be prefilled into the installer so you don’t need to worry about specifying that at any time during the installation process.

The Workaround Method

Before we had talked to our Microsoft account managers, this is the route we were going to use to be able to install SQL Server 2022 on our VMs. Go to the SQL Server 2022 page on the Microsoft website (https://www.microsoft.com/en-us/sql-server/sql-server-downloads) then download the “Developer Edition”.

After you download that version, run the installer that you downloaded. When it first opens, click the option that says “Download Media”:

On the next page, click “Download”, which will download the ISO for SQL Server:

After the ISO file has downloaded, double-click on the file in File Explorer, which will mount a “DVD” to your system, which will open another directory. In that directory, double-click on “setup.exe” to run the installer for SQL Server.

When the installer window opens, click on the “Installation” page from the left-hand pane, then click on the first link that says “New SQL Server standalone installation or add features to an existing installation”.

If you have an issue where clicking that link doesn’t open anything else, seems to do nothing, check out my other post about how to resolve that issue.

Click “Next” multiple times to move through the initial installation steps for SQL Server 2022. Review the information on each page before moving past it to ensure the settings are what you want them to be. When you come to the following page, keep the radio button for “Perform New Installation” selected then click “Next” again.

The “Edition” page is the one we’ve been needing to get to. On this page, you can change the selection from “Specify a free edition” to either one of the other two options to install Standard or Enterprise editions.

If you select the option for “Use Pay-As-You-Go billing through Microsoft Azure”, there is a dropdown menu that will let you specify Standard or Enterprise editions. Note that going with this option would then later require you to specify Azure Subscription information to bill the server usage to.

If you select the option for “Enter the product key”, you can then give a license key for SQL Server 2022 if you already have one. Review the two optional checkboxes under the license key box to select any options that apply to you. It looks like if you don’t specify a license key after choosing this option, you can still move forward with the installation, but there must be some point in the future of using the software where you’ll be required to provide a license key to keep using it.

Note: If you go through the “proper method” I listed above, the license key should be prefilled in the third option when you get to this page with the properly downloaded ISO file. For our situation, we also need to make sure the checkbox below the license key box “I have a SQL Server license with Software Assurance or SQL Software Subscription” is checked, since we use Software Assurance for our licensing.

After you’ve made your selections for the “Edition” page of the installer, you are ready to move through the rest of the installation process, specifying whatever options required for your own version of SQL Server.

Summary

There are two different methods I have found for getting the installer for SQL Server 2022 Standard edition, and neither of them are necessarily intuitive. Based on your needs, you can choose either the “proper” or the “workaround” methods. I hope this tutorial has helped you download the software without having to spend a day figuring it out like my coworkers and I had to.

How to Get Role Assignments for a Resource in Azure Using Python

If you are in a role similar to one I am in now, where there is heavy oversight into certain data the company owns due to federal regulations, you may get assigned a similar project where you need to automatically retrieve various ownership and access information about one or more databases in your Azure cloud environment. What I’ve found through my own process of completing such a project is that the documentation Microsoft offers for its Python SDKs is abysmally low, and there is often nothing helpful being offered about the class you are trying to work with. But I have finally figured it out with my own determination, trial, and effort so I decided I would share it with everyone so that hopefully others don’t have to go through the same amount of effort to figure out something that should be fairly easy to accomplish.

What’s In This Post

The Task

Get a list of the “role assignments” for a SQL Server resource in the Azure Portal, specifically the Owner assignments. According to this Microsoft article, the “Owner” role “grants full access to manage all resources, including the ability to assign roles in Azure RBAC.” Since that is a lot of freedom, our auditors want to make sure they know who has such access so they can determine if those people should have access.

Necessary Python Modules

There are two modules that you will need to download and install in order to programmatically generate the list of “Owner” assignments on a resource in Azure. The first is from azure.identity import ClientSecretCredential, which is how we are going to create credentials to use with the Authorization Management client. The second is from azure.mgmt.authorization import AuthorizationManagementClient, which is the module that will allow us to query the Authorization API that Microsoft provides, so that we can get that list of role assignments for a particular resource in the cloud environment.

Required Permissions

In order for you to be able to query this data we are looking for, you will need to create an app registration that has been given the appropriate permissions on either the subscription or resource group level, depending on the security requirements of your organization. Lowest access is always preferable, but giving the permission at the subscription level would then allow your app registration to perform this same action on any and all resources you want with just one permission assignment. The permission you need to assign to your app registration at either the Resource Group or Subscription level is “API Management Service Reader Role”, which can be assigned under “Access Control (IAM)” for the Subscription or Resource Group.

Data Pieces You Will Need

There are several pieces of information you will need to know and feed in to your Python script to get it to retrieve the role assignments for a given resource.

  • Tenant ID, Client ID, and Client Secret of the app registration you are using with your script
    • These three items are used to generate the credential needed for the Authorization API using the Azure Identity ClientSecretCredential
    • To keep these vital pieces of information secure, you can review my upcoming blog post about encrypting your IDs and keeping your secrets safe while still using them
  • Subscription ID for the subscription your resource is in, even if you are getting the role assignments at the resource group or resource level (not Subscription level)
  • Resource Group name for the group your resource is in
  • Resource Namespace. This item is a bit more confusing to achieve since there really isn’t any clear information about what it means, at least in my own research. (If you can find a Microsoft document talking about this, please let me know!) What I found through my own trial and error is that this value, at least if the resource you’re getting the owners for is a SQL Server resource, will be "Microsoft.Sql". I ultimately figured this out by looking at the full Resource ID for one of my SQL Server instances and seeing that value there. You can also use the Resource Explorer in the Azure portal and look in the “Providers” list to see that there are hundreds of different namespaces that can be used for getting the role assignments for any resource.
    • In this screenshot below, taken from the Properties page of my SQL Server resource, you can see the full ID for the resource, which includes "Microsoft.Sql/servers" which is the namespace/resource type for this resource
Screenshot showing the full Resource ID for a SQL Server resource, which includes the Resource Namespace and the Resource Type
  • Resource Type. Similarly to getting the resource namespace, it’s hard to find the official value that should be used for Resource Type when querying the Authorization Management API. But I figured it out from the Resource ID like I did the namespace. The Resource Type can also be found under the Resource Explorer “Providers” list, but just looking at that did not clarify for me what value I should be passing in. I once again found out through trial and error what value would work for this parameter.
  • Resource Name. The name of the resource you want to get role assignments for. If you have a SQL Server called “prod_dta”, then you would pass that value in for this parameter.
  • Optional: The ID value for the role type you are looking for, which in my case, was the Owner role. This is a standard ID that is assigned to the Owner role across your organization (or maybe it’s the same for all of Microsoft…). See subsection below for how to find this ID value.

Getting a Role ID Value

In the Azure portal, navigate to any resource and then go to the “Access Control (IAM)” page:

In the IAM page, click on the “Roles” tab then click on “View” for the Owner role (or whichever role you would like to get the ID for)

In the pane that opens with the details of the role, you want to click on the “JSON” tab, which is where you will find the ID you need. The full ID of the role is longer than just the GUID we are looking for, so make sure to only copy out the GUID from the end of the string (which is what’s covered in gray on the screenshot below). Save this ID for later.

The Code

Once you have figured out and gathered all the required pieces of data needed to get role assignments for a resource in Azure, the code for getting that data is fairly simple. Below is all of my code which I used to get just the people with Owner role assignments from a resource.

from azure.identity import ClientSecretCredential
from azure.mgmt.authorization import AuthorizationManagementClient
import json


def create_auth_management_client(tenantID, clientID, clientSecret, subscriptionID):
    credentials = ClientSecretCredential(tenant_id=tenantID, client_id=clientID, client_secret=clientSecret)
    authMgmtClient = AuthorizationManagementClient(credential=credentials, subscription_id=subscriptionID)


def list_resource_owners(authMgmtClient, serverName):
    resourceGroupName = "MyResourceGroup"
    resourceNamespace = "Microsoft.Sql"
    resourceType = "servers"
    resourceName = serverName
    ownerRoleDefID = "f865f414-b2bf-4993-8f94-51b055da4356" 
    # This is a random GUID value that I replaced with the actual GUID for the Owner role in our system. Since I'm not sure if those are unique to organizations or not, I have taken out the actual value for this example.
    
    permissions = authMgmtClient.role_assignments.list_for_resource(resource_group_name=resourceGroupName, resource_provider_namespace=resourceNamespace, resource_type=resourceType, resource_name=resourceName)
    principalList = []

    for item in permissions:
        stringItem = str(item)
        if ownerRoleDefID in stringItem: #If the current item is for an Owner role
            # Replace single quotes with double quotes in string
            replacedStringItem = stringItem.replace("'","\"")

            # Replace None with "None"
            replacedStringItem = replacedStringItem.replace("None","\"None\"")

            # Get index at which to cut off the string
            stopIndex = replacedStringItem.find(", \"principal_type\"")

            # Substring the string item to stop after getting the princiapl id
            substring = replacedStringItem[0:stopIndex]

            # Add closing bracket to end of the substring
            substring = substring + '}'
            
            # Load the item into a JSON-format object
            jsonItem = json.loads(substring)

            # Get just the principal ID value from the JSON object
            principalID = jsonItem['principal_id']

            # Add the principal ID to the principalList object
            principalList.append(principalID)
    return principalList

Let’s break that down further

from azure.identity import ClientSecretCredential
from azure.mgmt.authorization import AuthorizationManagementClient
import json

The above code snippet shows the 3 libraries you need to import for this code to work.

  1. ClientSecretCredential from Azure.Identity: This module/library allows you to create a login credential for your Azure environment.
  2. AuthorizationManagementClient from azure.mgmt.authorization: This library is what gives you the modules, functions, and classes needed to interact with role assignments in Azure, which includes functions to create, update, and delete role assignments with your Python code. I only used the “get” functionality because I only wanted to retrieve assignments, not make new ones.
  3. JSON: I use this library to manipulate the data returned from the Auth Management API calls in a JSON format, which is easier to work with to retrieve the value associated with a particular key.
def create_auth_management_client(tenantID, clientID, clientSecret, subscriptionID):
    credentials = ClientSecretCredential(tenant_id=tenantID, client_id=clientID, client_secret=clientSecret)
    authMgmtClient = AuthorizationManagementClient(credential=credentials, subscription_id=subscriptionID)

This function, create_auth_management_client takes input of the Tenant ID, Client ID, and Client Secret of the app registration making the calls to Authorization Management, as well as the Subscription ID that the resource we want to get role assignments for is in, and uses the ClientSecretCredential function from the Microsoft library to generate such a credential. It then makes a client to interact with the Authorization Management API with Azure using that credential. The authMgmtClient object will be the object we call methods on to get the data we need.

We then come to the definition of the function list_resource_owners, which is where I pull the role assignment data from Azure, then parse the data returned into a usable format and pull out just the role assignments that contain the Owner role ID value. The beginning of the function first sets important data values for the Authorization Management Client to use; you could pass these in to your own function if you want instead of hard-coding the values within the function.

    resourceGroupName = "MyResourceGroup"
    resourceNamespace = "Microsoft.Sql"
    resourceType = "servers"
    resourceName = serverName  # Passed in as parameter to function
    ownerRoleDefID = "f865f414-b2bf-4993-8f94-51b055da4356" # Replace with correct value from your environment

After defining those important values, the code then finally makes a call to Authorization Management:

permissions = authMgmtClient.role_assignments.list_for_resource(resource_group_name=resourceGroupName, resource_provider_namespace=resourceNamespace, resource_type=resourceType, resource_name=resourceName)

If you were to then print(permissions), you would get a very long string of role assignments for the given resource (but permissions is an Authorization Management object, not a true string). The results look something like this (except it would contain the GUIDs of all the objects, not the all-zero GUIDs I redacted with):

{'additional_properties': {}, 'id': '/providers/Microsoft.Management/managementGroups/00000000-0000-0000-0000-000000000000/providers/Microsoft.Authorization/roleAssignments/00000000-0000-0000-0000-000000000000', 'name': '00000000-0000-0000-0000-000000000000', 'type': 'Microsoft.Authorization/roleAssignments', 'scope': '/providers/Microsoft.Management/managementGroups/00000000-0000-0000-0000-000000000000', 'role_definition_id': '/subscriptions/00000000-0000-0000-0000-000000000000/providers/Microsoft.Authorization/roleDefinitions/00000000-0000-0000-0000-000000000000', 'principal_id': '00000000-0000-0000-0000-000000000000', 'principal_type': 'User', 'description': None, 'condition': "((!(ActionMatches{'Microsoft.Authorization/roleAssignments/write'})) OR (@Request[Microsoft.Authorization/roleAssignments:RoleDefinitionId] ForAnyOfAllValues:GuidNotEquals {00000000-0000-0000-0000-000000000000, 00000000-0000-0000-0000-000000000000, 00000000-0000-0000-0000-000000000000})) AND ((!(ActionMatches{'Microsoft.Authorization/roleAssignments/delete'})) OR (@Resource[Microsoft.Authorization/roleAssignments:RoleDefinitionId] ForAnyOfAllValues:GuidNotEquals {00000000-0000-0000-0000-000000000000, 00000000-0000-0000-0000-000000000000, 00000000-0000-0000-0000-000000000000}))", 'condition_version': '2.0', 'created_on': datetime.datetime(2024, 6, 14, 21, 30, 43, 130165, tzinfo=<isodate.tzinfo.Utc object at 0x0000020C0D71D490>), 'updated_on': datetime.datetime(2024, 6, 14, 21, 30, 43, 130165, tzinfo=<isodate.tzinfo.Utc object at 0x0000020C0D71D490>), 'created_by': '00000000-0000-0000-0000-000000000000', 'updated_by': '00000000-0000-0000-0000-000000000000', 'delegated_managed_identity_resource_id': None}

You would have one of these sections of data for every single role assignment on the specified resource.

An important note is that there are several different classes you can call a list_for_resource function on from the authMgmtClient, but the only one that gets this data for the entire resource, and not just for the role assignments that apply to the caller, you need to use the role_assignments class. Here is a link to the Microsoft documentation, it honestly wasn’t that useful to me, but may be useful to you.

The final step of the processing of role assignments is to format it into something that is more useful, a JSON object, which is what the following section of code aims to do.

    principalList = []

    for item in permissions:
        stringItem = str(item)
        if ownerRoleDefID in stringItem: #If the current item is for an Owner role
            # Replace single quotes with double quotes in string
            replacedStringItem = stringItem.replace("'","\"")

            # Replace None with "None"
            replacedStringItem = replacedStringItem.replace("None","\"None\"")

            # Get index at which to cut off the string
            stopIndex = replacedStringItem.find(", \"principal_type\"")

            # Substring the string item to stop after getting the princiapl id
            substring = replacedStringItem[0:stopIndex]

            # Add closing bracket to end of the substring
            substring = substring + '}'
            
            # Load the item into a JSON-format object
            jsonItem = json.loads(substring)

            # Get just the principal ID value from the JSON object
            principalID = jsonItem['principal_id']

            # Add the principal ID to the principalList object
            principalList.append(principalID)
    return principalList

What that above code is doing is that for every item in the returned permissions object (each line like the example I showed above), formatting is going to be applied.

  1. The current item will be converted to a true string object using str()
  2. If the current item contains the ID of the owner role (which we retrieved earlier), then we will keep processing it. If it doesn’t contain that ID, then the item will be ignored and processing started on the next item in the object.
  3. The first formatting change is to replace any single-quote characters with double-quote characters, which is what JSON expects. Example: 'name': '00000000-0000-0000-0000-000000000000' will change to "name": "00000000-0000-0000-0000-000000000000"
  4. The next formatting change is to replace any instances of None with "None", which also better fits JSON formatting.
  5. Because of weird formatting that happens later in each line of the permissions object, I opted to cut off the string after the final piece of data that I needed, which was the “principal_id” value. So the line stopIndex = replacedStringItem.find(", \"principal_type\"") is retrieving the character index at which the string “, principal_type” is found, so that I can cut off the string right before that value.
  6. The next step of formatting is to perform a string slice on the current item to cut it off at the designated point. That means the above example data will now look like {'additional_properties': {}, 'id': '/providers/Microsoft.Management/managementGroups/00000000-0000-0000-0000-000000000000/providers/Microsoft.Authorization/roleAssignments/00000000-0000-0000-0000-000000000000', 'name': '00000000-0000-0000-0000-000000000000', 'type': 'Microsoft.Authorization/roleAssignments', 'scope': '/providers/Microsoft.Management/managementGroups/00000000-0000-0000-0000-000000000000', 'role_definition_id': '/subscriptions/00000000-0000-0000-0000-000000000000/providers/Microsoft.Authorization/roleDefinitions/00000000-0000-0000-0000-000000000000', 'principal_id': '00000000-0000-0000-0000-000000000000'
  7. Then to return that string to standard JSON formatting, we are going to append a closing curly bracket to the string
  8. After getting the string formatted nicely so that it can be a JSON object, I use json.loads(substring) to create a true JSON object from that string
  9. Once the string has been turned into a JSON object, I use standard JSON string extraction formatting (principalID = jsonItem['principal_id']) to pull just the principal_id value out of the JSON object. If I had not gone this route, it would have been much more difficult to do string parsing to correctly get the value of the principal_id key.
  10. The final step of this function is to add that retrieved principal_id value to the principalList List and then return that List to the caller once all lines of the permissions object have been extracted, formatted, and the principal_id values retrieved.

Summary

There isn’t much documentation and there aren’t many references online about how to retrieve role assignment data for a resource from Azure, so I hope this tutorial has been helpful to you. My goal for this post was to make it so that others did not have to go through all the trial and error I had to go through to figure this out. I think that once you get an example of the data that the API is expecting you to pass in, it becomes much easier to query not only SQL Server resources but also any other type of resources in Azure. Let me know in the comments below if you found this tutorial helpful or if you have any remaining questions or misunderstandings on the topic that you would like help with.

First Thoughts About Azure

As many of you probably already know, my cloud development career started in AWS, which I worked with for just about 3 years while I worked at Scentsy. Since my recent transition to a new job at a different company, I have started to develop in Azure instead, and it’s been a learning journey. Although both platforms allow for cloud development and processing, they have quite a few notable differences in what is offered and how they offer it, which is what I’m going to cover in this post today. My goal for this list isn’t to have a technical or all-inclusive list of the differences, but more of a difference a developer might feel in their own work if they make the same switch that I have.

What’s in this post:

Azure seems simpler

Azure is simpler yet still robust. Sometimes I feel like AWS tries to overcomplicate their services in order to make them seem fancier or more cutting-edge. And it also seems like they split what could be one service into multiple just to increase their total service count. Azure combines multiple functions I was used to in AWS into a single service. An example of that is Azure DevOps, which combines your ticketing/user story system with your DevOps pipelines and your Git (or other) repos. In my past job, we used TeamCity and Octopus Deploy for the pipelines, Jira for the ticketing, and Bitbucket to store our code, so I was a little confused my first couple of weeks in my new role since everything seemed to only be in one location. But I now find it nice and easier to work with.

Azure has better cloud ETL development

In the Azure cloud platform, there is a service called Synapse Workspace or Synapse Studio, and a second service called Azure Data Factory, which both allow you to create ETL pipelines right in the cloud. AWS has Glue, but that really doesn’t seem to have the same feel or capabilities that either Synapse or Azure Data Factory (ADF) has in the Azure realm. I have already updated and created several pipelines in each of those services in Azure and I really enjoyed working with them because they were very intuitive to get working with as a newbie and I could do everything I needed for the ETL right in the cloud development workspace.

When I worked with Glue in the past, it definitely did have some limited capabilities for making drag-and-drop ETLs in the cloud, but the service seemed to have a lot of limits which would force you to start writing custom PySpark code to make the data move. While writing custom code is also possible with Synapse and ADF, they both are built with more robust built-in components that allow you to make your ETLs quickly without writing any more custom code than a few SQL queries. I have really been enjoying working in these new services instead of AWS’ Glue.

More on Azure Data Factory

Another reason why I have been enjoying working with Azure Data Factory (ADF) is because it seems to be a modern version of the SSIS I am already familiar with, and located in the cloud instead of on an ETL server and local developer box. Although the look of ADF isn’t exactly the same as SSIS, it still is the drag-and-drop ETL development tool I love working with. And since it’s developed by Microsoft, you get all the best features available in SSIS ETL development without having to work with the old buggy software. I’m sure as I keep working with ADF that I’ll find new frustrating bugs that I’ll need to work around, but my experience with it so far has been only positive.

Power Automate & Logic Apps

Two other tools that aren’t available in the AWS ecosystem and that don’t seem to have an analog in AWS are Power Automate and Logic Apps. While these tools are more aimed at people who are not developers, to allow them to automate some of their daily work, they are interesting and useful features for certain scenarios and I am enjoying learning about them and playing with them. One of the best parts about working with Azure services is that it’s fully integrated into the entire Microsoft ecosystem, so you can pull in other non-Azure Microsoft services to work with Azure and expand your horizons for development. I’m not sure yet that I would 100% recommend working with Power Automate or Logic Apps for task automation (I’m still not done learning it and working with it), but it at least is another option to fall back on in the Microsoft realm that isn’t available in AWS.

Copilot isn’t what they want it to be

While most of my experience with Azure so far is positive, there are a couple annoying things I’ve noticed that I think are worth sharing, although neither of them are so egregious that it would prevent me from recommending working with this platform.

The biggest negative about Azure for me so far is that Microsoft keeps trying to shove Copilot (their AI assistance tool which seems only slightly more advanced than Clippy) into every single product they offer even when it provides no benefit or actually detracts from your total productivity. The perfect example of this is the “New Designer” for Power Automate. For some unknown reason, Microsoft has decided that instead of allowing you to do a drag-and-drop interface for task components to build your automation flow, everyone should instead be required to interact with Copilot and have it build your components instead. That might be useful if you had already been working with Power Automate in the past so knew what capabilities and components it offered. But as someone totally new to this space who is trying to learn how to use the tool and has no idea what is currently possible to develop, it feels basically impossible to communicate with the AI in any meaningful way in order to build what I want. I don’t know what to ask it to create when I’ve never seen a list of tasks that are available. Luckily, for now it is possible to toggle off the “New Designer” and switch back to the old that allows you to add each individual component as you go and select those components from a list which gives you a short description of what each does. Maybe in the future I’ll be more open to using Copilot with everything I develop, but right now, as a new developer in Azure, it doesn’t work for me.

Unintuitive service naming

The only other nitpick I have about the Azure and Microsoft cloud ecosystem is that sometimes, the names they pick for their services don’t make sense, are confusing, or are the same thing as a totally different service. Microsoft doesn’t seem to be that great at naming things to make them understandable at a quick glance, but I suppose that can also be attributed to the desire of all cloud computing companies to make themselves look modern and cutting-edge.

The best example I can give of this phenomenon right now is that a data lake in Azure is built on what are called Storage Accounts, which is the blob storage service within Azure. It’s not as confusing to me now that I’ve been dealing with it for a month and a half, but that name doesn’t seem at all intuitive to me. Each time my colleagues directed me to go to the “data lake” I would get confused as to where I was supposed to navigate since the service I would click into was called Storage Accounts instead.

Summary

Although it felt like such a big switch in the beginning to move from an AWS shop to an Azure shop, I have already started to enjoy developing in Azure. It has so much to offer in terms of cloud ETL development and I can’t wait to keep learning and growing with these tools. I’ve already compiled so many things that I can’t wait to share, so I am hoping I will get those posts ready and posted soon so others can learn from my new Azure developer struggles.

Taking a Break and Switching Things Up

If you know me personally or follow me on LinkedIn, then you know that several weeks ago I made the decision to leave my friends and colleagues at Scentsy to take a new role at Boise Cascade. I made this decision for a lot of different reasons, but the main one being that I felt like I needed more of a challenge and wanted to get back into development work rather than focusing on project organization and management.

I am just about through my third week in my new role and it has been a scary, tiring, interesting, and even a bit fun, experience. Starting any new job comes with some new stress as I adjust to the environment and coworkers, and while I’m not completely out of the adjustment phase yet, I am already becoming more comfortable in this role so thought I should come here and give an update.

Previously, this blog has been heavily focused on cloud development with AWS since that is what I was developing in daily at Scentsy. But with my new role, I am drinking from a firehose to start learning cloud development in Azure, so my posts will be switching to focus on that platform instead. As of right now, I have no plans to do any personal development projects in AWS to be able to continue content in that space. I am really excited to be learning about Azure though, and am equally excited to start sharing what I learn about it here on my blog, since it seems like the Azure documentation is just as hard to understand, interpret and use as the AWS documentation is. So I will have plenty of my own learnings to share.

In my new role, I am also jumping into the deep end with scripting in Python which I am loving (but of course having issues with just like any other platform), so I will be starting to share some of my learnings about that space as well. Plus, there’s always a possibility I’ll be learning some other new technology because of how diverse my new role is, so my content going forward might be a lot more diverse than it was in the past.

It may take a few more weeks for me to get into the swing of things enough at my new job to feel like I have something worth posting about here, so I hope you’ll stick around and give the new content a read once it comes out. And as always, if there is anything in particular you would like me to write about, let me know in the comments and I will try to get to it!