refs

Author: brown9804

0_Azure/3_AzureAI/AIFoundry/demos/5_What-Global-DataZone-StdType.md

@@ -23,6 +23,10 @@ From [Enterprise trust in Azure OpenAI Service strengthened with Data Zones](htt
 
 > [!TIP]
 > The appropriate deployment model depends on your data residency and compliance requirements. Here's a breakdown of the three main options, and we also need to consider the different ways the models are available to be deployed:
+> `Global Deployments`: Suitable for scenarios where compliance with regional data laws is not required, and performance and scale are the primary concerns.
+>    -  Data `at rest and in transit are not restricted to any specific region or data zone`.
+>    -  Azure may `store or process data anywhere across its global infrastructure`.
+>    -  This model offers maximum flexibility and scalability, but provides the least control over data residency and movement.
 > - `Regional PTU`: Use this model when you have strict regional data residency requirements. For example, if you deploy in Sweden Central, all data (`both at rest and in transit`) remains within that `region’s boundaries.`
 > - `Standard Deployment`: Choose this when you need to keep data at rest within a specific Azure region, but are comfortable with some flexibility for data in transit.
 >   - You select the region (e.g., Sweden Central).
@@ -32,6 +36,25 @@ From [Enterprise trust in Azure OpenAI Service strengthened with Data Zones](htt
 >    - You select the `zone, not the region.`
 >    - Azure manages the region placement internally, ensuring all data remains within the selected zone.
 
+| **Deployment Type**             | **Scope**       | **Model Type**         | **Data Residency (At Rest & In Transit)**                                                                 | **Latency & Performance**                                                                 | **Use Case / Description**                                                                 |
+|--------------------------------|------------------|------------------------|------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------|---------------------------------------------------------------------------------------------|
+| Global Standard                | Global           | Standard               | Data at rest stays in the designated Azure geography; data in transit may be processed in any Azure region. | Low latency under normal usage; may experience **latency variation** at high volumes.     | Best for general use cases where flexibility and scale are prioritized over strict residency. |
+| Global Provisioned Managed     | Global           | Managed                | Same as Global Standard, but with **reserved model capacity** across global infrastructure.                | More consistent latency; suitable for **high-volume workloads** needing predictable performance. | Ideal for enterprise-grade deployments with consistent traffic and performance needs.       |
+| Global Batch                   | Global           | Batch                  | Same residency as other global types; data processed asynchronously across global regions.                 | **Up to 24-hour turnaround**; optimized for cost and throughput, not real-time latency.    | Large-scale processing like document summarization, content generation, or NLP tasks.       |
+| Data Zone Standard             | Data Zone        | Standard               | Data at rest stays in the designated geography; data in transit processed within the selected data zone.   | Similar to Global Standard; **latency may vary** at high volumes.                          | Ensures compliance with zone-level regulations (e.g., EU), with moderate performance control. |
+| Data Zone Provisioned Managed  | Data Zone        | Managed                | Same as Data Zone Standard, but with **reserved capacity** within the zone.                                | Lower latency variance; optimized for **predictable performance** in compliant environments. | Combines compliance with performance for regulated industries.                              |
+| Data Zone Batch                | Data Zone        | Batch                  | Same as other Data Zone types; batch processing within the zone.                                           | **Up to 24-hour turnaround**; cost-effective for non-real-time workloads.                  | Compliant batch processing for large datasets within a data zone.                           |
+| Regional PTU                  | Specific Region  | PTU-based              | Data at rest and in transit remain strictly within the selected Azure region.                              | Lowest latency variance; **tightest control** over data movement.                          | Required for strict regional compliance (e.g., financial or government workloads).           |
+| Standard Deployment           | Specific Region  | Standard               | Data at rest stays in region; data in transit may leave but stays within the same data zone.               | Moderate latency control; **some flexibility** in data routing.                            | Balanced option for regional control with some scalability.                                 |
+| Data Zone Deployment          | Data Zone        | Standard               | Both data at rest and in transit remain within the selected data zone; Azure chooses the region internally. | Similar to Data Zone Standard; **no control over specific region**, but zone compliance maintained. | Good for zone-level compliance without needing to manage region specifics.                  |
+
+> [!TIP]
+> - **Global Batch** and **Data Zone Batch**: Target turnaround time is **up to 24 hours**, making them unsuitable for real-time applications.
+> - **Standard deployments** (Global/Data Zone/Regional): May experience **latency variation** at high usage volumes.
+> - **Provisioned deployments**: Offer **reserved capacity**, reducing latency variance and improving predictability.
+> - **Streaming**: Can reduce perceived latency by returning tokens incrementally, especially useful in chat interfaces.
+> - **Content Filtering**: Adds safety but may increase latency slightly.
+
 <details>
 <summary><b>List of References</b> (Click to expand)</summary>