[Design Pattern] Handling race conditions and state in serverless data pipelines
Hello community, I recently faced a tricky data engineering challenge involving a lot of Parquet files (about 2 million records) that needed to be ingested, transformed, and split into different entities. The hard part wasn't the volume, but the logic. We needed to generate globally unique, sequential IDs for specific columns while keeping the execution time under two hours. We were restricted to using only Azure Functions, ADF, and Storage. This created a conflict: we needed parallel processing to meet the time limit, but parallel processing usually breaks sequential ID generation due to race conditions on the counters. I documented the three architecture patterns we tested to solve this: Sequential processing with ADF (Safe, but failed the 2-hour time limit). 2. Parallel processing with external locking/e-tags on Table Storage (Too complex and we still hit issues with inserts). 3. A "Fan-Out/Fan-In" pattern using Azure Durable Functions and Durable Entities. We ended up going with Durable Entities. Since they act as stateful actors, they allowed us to handle the ID counter state sequentially in memory while the heavy lifting (transformation) ran in parallel. It solved the race condition issue without killing performance. I wrote a detailed breakdown of the logic and trade-offs here if anyone is interested in the implementation details: https://medium.com/@yahiachames/data-ingestion-pipeline-a-data-engineers-dilemma-and-azure-solutions-7c4b36f11351 I am curious if others have used Durable Entities for this kind of ETL work, or if you usually rely on an external database sequence to handle ID generation in serverless setups? Thanks, Chameseddine
Azure SQL Database : Can I use same primary key column and foreign key column for multiple tables?
CREATE TABLE Table1( PRIMARY KEY (Table1ID), Column2 int ); CREATE TABLE Table2( PRIMARY KEY (Table1ID), Column2 int, FOREIGN KEY (Table1ID) REFERENCES Table1(Table1ID) ); CREATE TABLE Table3( PRIMARY KEY (Table1ID), Column2 int, FOREIGN KEY (Table1ID) REFERENCES Table1(Table1ID) );Azure Logic App workflow (Standard) Resubmit and Retry
Hello Experts, A workflow is scheduled to run daily at a specific time and retrieves data from different systems using REST API Calls (8-9). The data is then sent to another system through API calls using multiple child flows. We receive more than 1500 input data, and for each data, an API call needs to be made. During the API invocation process, there is a possibility of failure due to server errors (5xx) and client errors (4xx). To handle this, we have implemented a "Retry" mechanism with a fixed interval. However, there is still a chance of flow failure due to various reasons. Although there is a "Resubmit" feature available at the action level, I cannot apply it in this case because we are using multiple child workflows and the response is sent back from one flow to another. Is it necessary to utilize the "Resubmit" functionality? The Retry Functionality has been developed to handle any Server API errors (5xx) that may occur with Connectors (both Custom and Standard), including client API errors 408 and 429. In this specific scenario, it is reasonable to attempt retrying or resubmitting the API Call from the Azure Logic Apps workflow. Nevertheless, there are other situations where implementing the retry and resubmit logic would result in the same error outcome. Is it acceptable to proceed with the Retry functionality in this particular scenario? It would be highly appreciated if you could provide guidance on the appropriate methodology. Thanks -Sri
Deriving expiry days and remaining retention days for blobs through blob inventory
In managing data within Azure blob storage accounts and Azure data lake gen 2 storage accounts, organizations often encounter scenarios where blobs have been deleted but remain in a soft-deleted state. To calculate the remaining retention days for all such blobs across an entire storage account can be a critical requirement for customers seeking to optimize data management and ensure compliance with retention policies. Additionally, certain blobs may have an expiry time set, scheduling their deletion for a future date. To facilitate the identification and monitoring of these blobs and their respective expiry times, a custom query has been written to efficiently list and calculate expiry information, enabling users to proactively manage their storage resources. The expiry time for Azure blobs is set using the Set Blob Expiry operation. This feature is present in only Hierarchical namespace enabled storage accounts. We can set the expiry with below steps: i) Azure Storage action- About Azure Storage Actions - Azure Storage Actions | Microsoft Learn Storage action can be used to set blob expiry, share a high-level snippet for the operation below ii) REST API- Set Blob Expiry (REST API) - Azure Storage | Microsoft Learn to set the expiry time for your blobs. This ensures that each blob has a defined lifecycle and will be deleted after the specified period. In this blog, it is a step-by-step process of listing the expiry time and retention of the blobs using Blob Inventory report and then parsing it using Synapse. 1. Set blob inventory rule Get CSV file from blob inventory run. Go to the container where inventory reports are getting stored. Navigate to the recent date folder and get url of Blob Inventory csv life. Sharing the below snippet for reference: 2. Create an Azure Synapse workspace Next, create an Azure Synapse workspace where you will execute a SQL query to report the inventory results. Create the SQL query: After you create your Azure Synapse workspace, do the following steps. Navigate to https://web.azuresynapse.net. Select the Develop tab on the left edge. Select the large plus sign (+) to add an item. Select SQL script. 3. Use the sample query below to get the expiry time and remaining retention days of blob respectively select LEFT([Name], CHARINDEX('/', [Name]) - 1) AS Container, RIGHT([Name], LEN([Name])- CHARINDEX('/',[Name])) AS Blob, [Expiry-time] from OPENROWSET( bulk '<URL to your inventory CSV file>', format='csv', parser_version='2.0', header_row=true ) as Source For blobs which got deleted directly, you can calculate the remaining retention days since the data is present in soft deleted state and will be deleted permanently after the retention days complete. select LEFT([Name], CHARINDEX('/', [Name]) - 1) AS Container, RIGHT([Name], LEN([Name])- CHARINDEX('/',[Name])) AS Blob, [Expiry-time], RemainingRetentionDays from OPENROWSET( bulk '<URL to your inventory CSV file>', format='csv', parser_version='2.0', header_row=true ) as Source In the above snippet, the Null value represents that the blob is not deleted and no expiry time is set on the blob yet. Please Note: Calculating blob expiry from blob inventory is one way, customer can explore other options such as Powershell and Azure CLI to achieve the same. Reference links:- Set Blob Expiry (REST API) - Azure Storage | Microsoft Learn Create a storage task - Azure Storage Actions | Microsoft Learn Azure Storage blob inventory | Microsoft Learn Calculate blob count and size using Azure Storage inventory | Microsoft Learn
How to configure directory level permission for SFTP local user
SFTP is a feature which is supported for Azure Blob Storage with hierarchical namespace (ADLS Gen2 Storage Account). As documented, the permission system used by SFTP feature is different from normal permission system in Azure Storage Account. It’s using a form of identity management called local users. Normally the permission which user can set up on local users while creating them is on container level. But in real user case, it’s usual that user needs to configure multiple local users, and each local user only has permission on one specific directory. In this scenario, using ACLs (Access control lists) for local users will be a great solution. In this blog, we’ll set up an environment using ACLs for local users and see how it meets the above aim. Attention! As mentioned in Caution part of the document, the ACLs for local users are supported, but also still in preview. Please do not use this for your production environment. Preparation Before configuring local users and ACLs, the following things are already prepared: One ADLS Gen2 Storage Account. (In this example, it’s called zhangjerryadlsgen2) A container (testsftp) with two directories. (dir1 and dir2) One file uploaded into each directory. (test1.txt and test2.txt) The file system in this blog is like: Aim The aim is to have user1 which can only list files saved in dir1 and user2 which can only list files saved in dir2. Both of them should be unable to do any other operations in the matching directory (dir1 for user1 and dir2 for user2) and should be unable to do any operations in root directory and the other directory. Configuring local users From Azure Portal, it’s easy to enable SFTP feature and create local users. Here except user1 and user2, another additional user is also necessary. It will be used as the administrator to assign ACLs on user1 and user2. In this blog, it’s called admin. While creating the admin, its landing directory should be the root directory of the container and the permissions should be all given. While creating the user1 and user2, as the permission will be controlled by using ACLs, the containers and permissions should be left empty and the Allow ACL authorization should be checked. The landing directory should be configured to the directory which this user should have permission later. (In this blog, user1 should be on dir1 and user2 should be on dir2.) User1: User2: After local users are created, one more step which is needed before configuring ACL is to note down the user ID of user1 and user2. By clicking the created local user, a page as following to edit local user should show out and the user ID will be included there. In this blog, the user ID of user1 is 1002 and user ID of user2 is 1003. Configuring ACLs Before starting configuring ACLs, clarifying which permissions to assign is necessary. As explained in this document, the ACLs contains three different permissions: Read(R), Write(W) and Execute(X). And from the “Common scenarios related to ACL permissions” part of the same document, there is a table which contains most operations and their corresponding required permissions. Since the aim of this blog is to allow user1 only to list the dir1, according to table, we know that correct permission for user1 should be X on root directory, R and X on dir1. (For user2, it’s X on root directory, R and X on dir2). After clarifying the needed permissions, the next step is to assign ACLs. The first step is to connect to the Storage Account using SFTP as admin: (In this blog, the PowerShell session + OpenSSL is used but it’s not the only way. Users can also use any other way to build SFTP connection to the Storage Account.) Since assigning ACLs for local users is not possible to a specific user, and the owner of root directory is a built-in user which is controlled by Azure, the easiest way here is to give X permissions to all other users. (For concept of other users, please refer to this document) Next step is to assign R and X permission. But considering the same reason, it’s impossible to give R and X permissions for all other users again. Because if it’s done, user1 will also have R and X permissions on dir2, which does not match the aim. The best way here is to change the owner of the directory. Here we should change the owner of dir1 to user1 and dir2 to user2. (By this way, user1 will not have permission to touch dir2.) After above configurations, while connecting to the Storage Account by SFTP connection using user1 and user2, only listing file operation under corresponding directory is allowed. User1: User2: (The following test result proves that only list operation under /dir2 is allowed. All other operations will return permission denied or not found error.) About landing directory What will happen if all other configurations are correct but the landing directory is configured as root directory for user1 or user2? The answer to the above question is quite simple: The configuration will still work, but will impact the user experience. To show the the result of that case, one more local user called user3 with user ID 1005 is created but its landing directory is configured as admin, which is on root directory. The ACL permission assigned on it is same as user2 (change owner of dir2 to user3.) While connecting to the Storage Account by SFTP using user3, it will be landing on root directory. But per ACLs configuration, it only has permission to list files in dir2, hence the operations in root directory and dir1 are expected to fail. To apply further operation, user needs to add dir2/ in the command or cd dir2 at first.
Exclude Prefix in Azure Storage Action: Smarter Blob Management
Azure Storage Actions is a powerful platform for automating data management tasks across Blob and Data Lake Storage. Among its many features, Exclude Prefix stands out as a subtle yet critical capability that helps fine-tune task assignments. What Is the "Exclude Prefix" Feature? The Exclude Prefix option allows users to omit specific blobs or folders from being targeted by Azure Storage Actions. This is particularly useful when applying actions such as: Moving blobs to a cooler tier Deleting blobs Rehydrating archived blobs Triggering workflows based on blob changes For example, if you're running a task to archive blobs older than 30 days, but you want to exclude logs or config files, you can define a prefix like logs/ or config/ in the exclusion list. How to Use It — Example Scenario: In the following example, blobs across the entire storage account were deleted based on a condition: if a blob’s access tier was set to Hot, it was deleted except for those blobs or paths explicitly listed under the Exclude blob prefixes property. Create a Task: - Navigate to the Azure portal and search for Storage tasks. Then, under Services, click on Storage tasks – Azure Storage Actions On the Azure Storage Actions | Storage Tasks page, click Create to begin configuring a new task. Complete all the required fields, then click Next to proceed to the Conditions page. To configure blob deletion, add the following conditions on the Conditions page. Add the Assignment :- Click Add assignment in the Select scope section, choose your subscription and storage account, then provide a name for the assignment. In the Role assignment section, select Storage Blob Data Owner from the Role drop-down list to assign this role to the system-assigned managed identity of the storage task. In the Filter objects section, specify the Exclude Blob Prefix filter if you want to exclude specific blobs or folders from the task. In the example specified above, blobs will be deleted—except for those under the path “excludefiles” listed in the Exclude blobprefixes property. In the Trigger details section, choose the runs of the task and then select the container where you'd like to store the execution reports. Select Add. In the Tags tab, select Next and in the Review + Create tab, select Review + create. When the task is deployed, your deployment is complete page appears and select Go to resource to open the Overview page of the storage task. Enable the Task Assignment: - In the Trigger details section, we have a Enable task assignment checkbox which is checked by default. If the Enable task assignments checkbox is unchecked, you can still enable assignments manually from the Assignments page. To do this, go to Assignments, select the relevant assignment, and then click Enable. The task assignment is queued to run and will run at the specified time. Monitoring the runs:- After the task completes running, you can view the results of the run. With the Assignments page still open, select View task runs. Select the View report link to download a report. You can also view these comma-separated reports in the container that you specified when you configured the assignment. Conclusion: The Exclude Prefix feature in Azure Storage Actions provides enhanced control and flexibility when managing blob data at scale. By allowing you to exclude specific prefixes from actions like delete or tier changes, it helps you safeguard critical data, reduce mistakes, and fine-tune automation workflows. This targeted approach not only improves operational efficiency but also supports more granular data in Azure Blob Storage. Note:- Azure Storage Actions are generally available in the following public regions: https://learn.microsoft.com/en-us/azure/storage-actions/overview#supported-regions We can also exclude certain blobs using the “Not”operator when building task conditions. Blobs may be excluded based on specific blob or container attributes from the task conditions side as well—not just through task assignments. In the screenshot below, we are using the Not operator (!) to exclude blobs where the blob name is equal to "Test". Please refer: https://learn.microsoft.com/en-us/azure/storage-actions/storage-tasks/storage-task-conditions#multiple-clauses-in-a-condition. Reference Links:- About Azure Storage Actions - Azure Storage Actions | Microsoft Learn Storage task best practices - Azure Storage Actions | Microsoft Learn Known issues and limitations with storage tasks - Azure Storage Actions | Microsoft Learn
Granting List-Only permissions for users in Azure Storage Account using ABAC
In this blog, we’ll explore how to configure list-only permissions for specific users in Azure Storage, allowing them to view the structure of files and directories without accessing or downloading their contents. Granting list-only permissions to specific users for an Azure Storage container path allows them to list files and directories without reading or downloading their contents. While RBAC manages access at the container or account level, ABAC offers more granular control by leveraging attributes like resource metadata, user roles, or environmental factors, enabling customized access policies to meet specific requirements. Disclaimer: Please test this solution before implementing it for your critical data. Pre-Requisites: Azure Storage GPV2 / ADLS Gen 2 Storage account Make sure to have enough permissions(Microsoft.Authorization/roleAssignments/write permissions) to assign roles to users , such as Owner or User Access Administrator Note: If you want to grant list-only permission to a particular container, ensure that the permission is applied specifically to that container. This approach limits the scope of access to just the intended container and enhances security by minimizing unnecessary permissions. However, in this example, I am demonstrating how to implement this role for the entire storage account. This setup allows users to list files and directories across all containers within the storage account, which might be suitable for scenarios requiring broader access. Action: You can follow the steps below to create a Storage Blob Data Reader role with specific conditions using the Azure portal: Step 1: Sign-in to the Azure portal with your credentials. Go to the storage account where you could like the role to be implemented/ scoped to. Select Access Control (IAM)->Add-> Add role assignment: Step2: On the Roles tab, select (or search for) Storage Blob Data Reader and click Next. On the Members tab, select User, group, or service principal to assign the selected role to one or more Azure AD users, groups, or service principals. Click Select members. Find and select the users, groups, or service principals. You can type in the Select box to search the directory for display name or email address. Please select the user and continue with Step 3 to configure conditions. Step 3: The Storage Blob Data Reader provides access to list, read/download the blobs. However, we would need to add appropriate conditions to restrict the read/download operations. On the Conditions tab, click Add condition. The Add role assignment condition page appears: In the Add action section, click Add action. The Select an action pane appears. This pane is a filtered list of data actions based on the role assignment that will be the target of your condition. Check the box next to Read a blob, then click Select: Step 4: Add the build expression in such a way that the below expression evaluates to false, so that the result entirely depends on the above condition. Save On the Review + assign tab, click Review + assign to assign the role with the condition. After a few moments, the security principal is assigned the role. Please Note: Along with the above permission, I have given the user Reader permission at the storage account level. You could give the Reader permission at the resource level/resource group level/subscription level too. We mainly have Management Plane and Data Plane while providing permissions to the user. The Management plane consists of operation related to storage account such as getting the list of storage accounts in a subscription, retrieve storage account keys or regenerate the storage account keys, etc. The Data plane access refers to the access to read, write or delete data present inside the containers. For more info, please refer to: https://docs.microsoft.com/en-us/azure/role-based-access-control/role-definitions#management-and-dat... To understand about the Built-in roles available for Azure resources, please refer to: https://docs.microsoft.com/en-us/azure/role-based-access-control/built-in-roles Hence, it is important that you give minimum of ‘Reader’ role at the Management plane level to test it out in Azure Portal. Step 5: Test the condition (Ensure that the authentication method is set to Azure AD User Account and not Access key) User can list the blobs inside the container. Download/Read blob failed. Related documentations: What is Azure attribute-based access control (Azure ABAC)? | Microsoft Learn Azure built-in roles - Azure RBAC | Microsoft Learn Tutorial: Add a role assignment condition to restrict access to blobs using the Azure portal - Azure ABAC - Azure Storage | Microsoft Learn Add or edit Azure role assignment conditions using the Azure portal - Azure ABAC | Microsoft LearnWhy is the Azure Monitor chart showing dashed lines for the Availability metric?
Have you ever noticed the dash lines on the "Availability" Azure Monitor metric for your Storage Account? What's that about? Is your Storage Account down? Why is it that we see sections where there is a dashed line and other sections where there is a solid line? Well, it turns out that Azure metrics charts use dashed line style to indicate that there is a missing value (also known as “null value”) between two known time grain data points. Meaning, this behavior is by design, yes, by design. It is useful for identifying missing data points. The line chart is a superior choice for visualizing trends of high-density metrics but may be difficult to interpret for the metrics with sparse values, especially when corelating values with time grain is important. The dashed line makes reading of these charts easier but if your chart is still unclear, consider viewing your metrics with a different chart type. For example, a scattered plot chart for the same metric clearly shows each time grain by only visualizing a dot when there is a value and skipping the data point altogether when the value is missing, for my case this is what I get just after selecting the scatter chart: So, by using the scatter chart is a lot clearer now where are the missing data points. But why would there be any missing data point for the "Availability" Azure Monitor metric in the first place? Well, it turns out that the Azure Storage Resource Provider only reports the Availability data to Azure Monitor when there is ongoing activity, meaning, when the Storage Resource Provider is processing requests on any of its services (blob, queue, file, table). If you are not sending any requests to your Storage Account, you should expect the "Availability" Azure Monitor metric to show big sections of dashed lines and understand that this is by design. This is critical because if you don't know this, you may think that the following chart shows that your Storage Account has been "unavailable" for several hours: However, by taking a look at the scatter chart, we know now that there was a lot of inactivity on this Storage Account, and that at some point only 1 datapoint was showing "no availability", after which there was also an important gap again reflecting a period of no activity, for the Storage Account to then start receiving requests and for the "Availability" Azure Monitor metric to show 100% again: References ======== Chart shows dashed line https://docs.microsoft.com/en-us/azure/azure-monitor/essentials/metrics-troubleshoot#chart-shows-dashed-lineConverting Page or Append Blobs to Block Blobs with ADF
In certain scenarios, a storage account may contain a significant number of page blobs classified under the hot access tier that are infrequently accessed or retained solely for backup purposes. To optimise costs, it is desirable to transition these page blobs to the archive tier. However, as indicated in the following documentation - https://learn.microsoft.com/en-us/azure/storage/blobs/access-tiers-overview the ability to set the access tier is only available for block blobs; this functionality is not supported for append or page blobs. The Azure blob storage connector in Azure data factory is capable of copying blobs from block, append, or page blobs and copying data to only block blobs. https://learn.microsoft.com/en-us/azure/data-factory/connector-azure-blob-storage?tabs=data-factory#supported-capabilities Note: No extra configuration is required to set the blob type on the destination. By default, the ADF copy activity creates blobs as Block Blobs. In this blog, we will understand how to make use of Azure Data Factory to copy the page blobs to block blobs. Please note that this is applicable to append blobs as well. Let’s take a look at the steps ahead Step 1: Creating ADF instance Create an Azure data factory resource in the Azure portal referring to the following document - https://learn.microsoft.com/en-us/azure/data-factory/quickstart-create-data-factory After creation, click on "Launch Studio" as shown below Step 2: Creating datasets Create two datasets by navigating to Author -> Datasets -> New dataset. These datasets are used in source and sink for the ADF copy activity Select "Azure blob storage" -> click on continue -> select "binary" and continue Step 3: Creating Linked service Create a new linked service and provide the storage account name which contains page blobs Provide the file path where the page blobs are located. You would also need to create another dataset for destination. Repeat the steps from 3 to 6 to create another destination dataset to copy the blobs to the storage account as block blobs. Note: You can use same or different storage account for the destination dataset. Set it as per your requirements. Step 4: Configuring a Copy data pipeline Once the two datasets are created, now create a new pipeline and under "Move and Transform" section, drag and drop the "Copy data" activity as shown below. Under the Source and Sink sections from the drop down, select the source and destination datasets respectively which were created in the previous steps. Select the “Recursively” option and publish the changes. Source: Sink: Note: You can configure the filters and copy behaviour as per your requirements. Step 5: Debugging and validating Now as the configuration is completed, click on "Debug". If the pipeline activity ran successfully, you should be able to see "succeeded" status in the output section as below. Verify the blob type of the blobs in the destination storage account and it should show as block blob and access tier as Hot. After converting the blobs to block blobs, several methods are available to change their access tier to archive. These include implementing a blob lifecycle management policy, utilizing storage actions, or by using Az CLI or PowerShell scripts. Conclusion Utilising ADF enables the conversion of page or append blobs to block blobs, after which any standard method such as LCM policy or storage actions may be used to change the access tier to archive. This strategy offers a more streamlined and efficient solution compared to developing custom code or scripts. Reference links: https://learn.microsoft.com/en-us/azure/storage/blobs/access-tiers-overview https://learn.microsoft.com/en-us/azure/data-factory/connector-azure-blob-storage?tabs=data-factory#supported-capabilities https://learn.microsoft.com/en-us/azure/data-factory/copy-activity-overview https://learn.microsoft.com/en-us/azure/storage-actions/storage-tasks/storage-task-quickstart-portal https://learn.microsoft.com/en-us/azure/storage/blobs/lifecycle-management-overview https://learn.microsoft.com/en-us/azure/storage/blobs/archive-blob?tabs=azure-powershell#bulk-archive
