Some of the most important services in Azure have one thing in common: until now, they only work with object data hosted in Azure Blob Storage.
Microsoft Foundry. Microsoft Fabric. Azure Machine Learning. Azure Databricks. These services read and write data through object storage, the only door they recognize. If your data lived anywhere else, including on a file system, they couldn't see it, use it, or derive value from it.
Customers who needed the insight and innovation that Microsoft Fabric or Microsoft Foundry have to offer did what they had to do. They made less-than-ideal protocol workarounds, or they made copies.
The copy is the cost
Copying data into a Blob container so that an Azure service can see it creates a second dataset to pay for, a second dataset to secure, a sync window during which the two copies disagree, and the possibility of synchronization conflicts. For large datasets, it also means waiting for the entire copy to finish before any of the work can start.
None of that is the work you wanted to do, but it's the price you pay to connect your file data to the services you need.
Blob Access Points to the rescue
To address this issue, we are announcing that Qumulo is an Azure Blob Access Points launch partner. Every ANQ instance can now be treated, natively, as a trusted Blob access point. That means any Azure service that has only been able to talk to Azure Blob Storage can now talk natively and directly to ANQ, without protocol workarounds or making additional copies.
Microsoft Foundry and Microsoft Fabric are the big ones, but any cloud-native service that runs on Azure and interacts with object/blob storage can read data from your ANQ cluster. Point them at ANQ's native Blob access, and they can share data in real time with your SMB and NFS applications and workflows, all from the same unified namespace.
To be clear about the scope: your SMB and NFS clients can read and write the shared dataset, and their updates are visible to the Azure services in real time. The Azure services' own access to that data is read-only for now.
There is no migration window, no second dataset, and no new governance model to manage. All your file data stays where it is, and all the Azure services now connect directly to it.
What this means in practice
Your data stays home. The file system is always the one place where the data lives and the one platform from which you manage and share it.
Azure's AI and analytics services work from the same live data as your file workloads. Foundry, Fabric, Azure ML and other services see the current state of your data, not a snapshot from the last sync.
Less to run. There are no copy jobs to build, schedule, or monitor, and no duplicate storage bill.
Part of a bigger story
Blob Access Points work hand in hand with the Qumulo Blob Accelerator. The Accelerator attaches to a Blob container you already own and leaves your data in place. Blob Access Points go the other way, letting Azure services reach the file data that lives in Qumulo. Together, they remove the need to copy data in either direction: data on your ANQ file system is visible in real time to your object-based workflows that rely on Azure Blob storage, while file-based applications can read your Azure Blob Storage data over industry-standard NFS and SMB protocols. Read more in our post on the Qumulo Blob Accelerator.
The question used to be "how do we get our data into the format this service needs?" Now it's "which Azure service do we want to point at it?"