Dynamics 365 Finance: Parallel Dataverse LTR Archive Jobs, Version Requirements and 3/5 Concurrency Limits

Under Microsoft’s AI at Work roadmap (ID 576773) and the broader One Dynamics One Platform cross-app initiative, that single-file bottleneck has been replaced by a multi-lane highway. Microsoft has rolled out parallel processing for Archive long-term retention job execution across Worldwide (Standard Multi-Tenant) cloud instances. According to Microsoft’s release plan documentation on Microsoft Learn, the capability entered public preview on July 27, 2026, following an earlier private preview announced on the Microsoft Dynamics 365 Blog in March 2026, and reached general availability on August 6, 2026.

While the headline promise is higher throughput and shorter overall archive durations, the actual engineering under the hood comes with strict version prerequisites, non-negotiable concurrency caps, and specific behaviors for existing job queues that every Dynamics 365 administrator and technical architect needs to factor into their maintenance windows.

Breaking the Sequential Bottleneck: What Changed​

In Dynamics 365 Finance and Operations, the archival framework lets organizations move inactive historical records that must be kept for audit, legal, or regulatory compliance out of hot, expensive live application tables and into Dataverse long-term retention. According to Microsoft Learn documentation, standard supported scenarios include:

• Dynamics 365 Finance: General ledger and Tax transactions
• Dynamics 365 Supply Chain Management: Inventory transactions, Inventory journals, and Sales orders
• Dynamics 365 Commerce: Commerce transactions

Previously, every scheduled archive job executed sequentially in the order it was created. With the parallel execution upgrade, each archive job still adheres to the core architectural boundary of being defined per legal entity and per contract or scenario type (such as General Ledger, Tax Transactions, or Sales Orders). However, the execution pipeline is now partitioned across job creation, marking, and scheduling so that eligible jobs can run concurrently.

Why does this matter in the real world? Consider a multinational enterprise closing out its fiscal year across two dozen subsidiaries. Under the legacy model, kicking off a General Ledger or Sales Order archive job for a massive North American legal entity meant European and Asian subsidiaries had to wait in line until that job completed its scheduling and execution phases. By partitioning the workload by legal entity, multiple partitions can now process at the same time—even within the exact same functional scenario.

Section Summary:

• Dynamics 365 Finance and Operations now supports concurrent execution for Dataverse long-term retention archive jobs.
• Jobs remain scoped per legal entity and contract type, while partitioning across job creation, marking, and scheduling allows simultaneous processing.
• The feature entered public preview in July 2026 and became generally available in August 2026.

Version Prerequisites and Default Enablement​

Before you start planning a weekend archive blitz, you need to verify both your Finance and Operations application build and your Power Platform solution levels. While the roadmap lists the feature as enabled by admins, makers, or analysts for users, Microsoft’s technical configuration guide on Microsoft Learn clarifies that parallel processing turns on automatically by default as soon as an environment satisfies two mandatory prerequisites. There is no separate UI toggle to flip once you are on the required bits.

First, your environment must run one of the following minimum Finance and Operations application builds:

Second, administrators must open the Power Platform admin center (PPAC) and update the Finance and operations virtual entity solution to version 2.10.0.9 or later. Missing either piece of this puzzle means your environment will remain stuck in the old sequential execution mode.

Section Summary:

• Parallel archive processing requires specific minimum builds across PU69 through PU72 (10.0.45 to 10.0.48) plus version 2.10.0.9 or later of the Finance and operations virtual entity solution in PPAC.
• Once both requirements are satisfied, the feature is enabled by default without requiring a manual feature toggle.

How the Partition Scheduler and Hard Concurrency Caps Work​

Allowing unthrottled parallel database operations on a live ERP system would be a fast track to starving production users of SQL resources and burning through Power Platform API capacity. To prevent archiving from disrupting daily business operations, Microsoft engineered the parallel scheduler around strict partitioning and fixed concurrency ceilings.

According to Microsoft Learn, the scheduler partitions jobs by legal entity and processes them in first-in, first-out (FIFO) order within each scenario's scheduled job group based on creation time. Jobs created earlier are always started before jobs created later.

To keep system load predictable across environments of all sizes, Microsoft enforces two hard system defaults:

1. Maximum concurrent jobs per scenario type: 3
2. Maximum concurrent jobs across all scenario types (global limit): 5

What does that look like in practice? You can run three General Ledger archive jobs for three different legal entities simultaneously, alongside two Sales Order archive jobs, hitting the global ceiling of five active jobs. Because of the three-per-scenario cap, a massive backlog of General Ledger jobs cannot monopolize all five global slots and starve out Inventory or Tax transaction archiving.

Can an administrator raise those limits if they have a quiet holiday weekend and want to push ten jobs at once? Right now, the answer is no. Microsoft explicitly notes in its documentation that these concurrency limits cannot be modified through the user interface or standard self-service configuration, though the engineering team is evaluating ways to offer configurable limits in a future release.

Section Summary:

• Jobs within a scenario group are scheduled in FIFO order based on creation time.
• System defaults cap concurrent execution at 3 jobs per scenario and 5 jobs globally across all scenarios to protect database health and Power Platform API capacity.
• Administrators cannot currently override or raise these limits through self-service settings.

The "Existing Queue" Catch and Lifecycle Boundaries​

Even after your environment is updated and parallel processing activates, you might notice your current queue still inching along one job at a time. That is by design, and it catches plenty of administrators off guard.

According to Microsoft Learn, only archive jobs created after parallel processing is enabled can run in parallel. Why? The parallel scheduler relies on a legal-entity partition key assigned at the exact moment a job is created. Any jobs that were already created and sitting in the queue prior to enablement lack this partition key and cannot be retroactively enrolled into parallel execution.

When the feature first activates in your environment, two rules govern the transition:

• Legacy queued jobs drain first: Pre-existing jobs must finish sequentially in the order they were originally scheduled. If you have dozens of old jobs waiting, draining that backlog could take considerable time—unless an administrator steps in and cancels the pending jobs that still sit in a Scheduled status and recreates them.
• New jobs run concurrently: Any newly created jobs after enablement receive the required partition key and enter the parallel scheduling pipeline.

It is also vital to understand which parts of the archival lifecycle actually speed up. When an archive job runs in Finance and Operations, it moves through five distinct stages:

1. Replicating data from live application tables to Dataverse long-term retention.
2. Marking qualifying records as ready for archiving in the live tables.
3. Marking those records as retained (archived) in Dataverse LTR.
4. Running a reconciliation check to verify all marked live records exist in Dataverse LTR.
5. Moving the marked records from live application tables into Finance and Operations history tables (and deleting them from the live tables).

Microsoft’s documentation clarifies that parallel processing does not affect the move-to-history phase. That final phase runs independently as part of each individual job’s normal lifecycle. Furthermore, Microsoft’s core archiving guide notes that the overall multi-stage background process for Dataverse LTR can still take up to 14 days end-to-end, and automatic database tuning can take up to seven days to reflect reduced history-table capacity (and up to a day for <tablename>-Retained tables to reflect in Dataverse capacity reports). While Dataverse LTR compresses live data to consume roughly 50 percent less database capacity on average—and unindexed history tables save roughly 30 percent or more compared to live tables—parallel job execution accelerates how many jobs can be scheduled and processed concurrently, rather than eliminating the background lifecycle stages.

Section Summary:

• Archive jobs queued before the feature was enabled lack partition keys and will continue to run one at a time until the old queue drains or admins cancel and recreate scheduled jobs.
• The move-to-history stage of each job still runs independently and is unaffected by parallel scheduling.

Guidance for Custom Scenarios and Workspace Monitoring​

What if your organization or implementation partner built custom archive scenarios for bespoke tables? The parallel framework supports custom scenarios, but only if developers explicitly wire them up.

According to Microsoft Learn and the Microsoft Dynamics 365 Blog, custom archive scenarios must assign a job criteria key to every job they create (such as calling setJobCriteriaKey() on the builder with a logical partition identifier, most commonly the legal entity or DataAreaId ). Two concurrent jobs under the same scenario with different criteria keys must target completely disjoint, non-overlapping record sets. If a custom scenario omits the job criteria key, it won't break—it will simply fall back to running its jobs sequentially, one at a time.

Once your jobs are running, administrators can track concurrent execution directly in the Archive with Dataverse long term retention workspace inside Dynamics 365 Finance and Operations. As highlighted on the Microsoft Dynamics 365 Blog, each running job displays its specific criteria key value in the workspace, allowing IT teams to visually confirm which legal-entity partitions are processing simultaneously.

Actionable Checklist for Dynamics 365 Administrators​

• Audit your versions: Verify that your Finance and Operations environment meets the minimum build for PU69–PU72 and update the Finance and operations virtual entity solution to 2.10.0.9 or higher in the Power Platform admin center.
• Clear or recreate old queues: Check for long queues of pre-existing jobs in Scheduled status; cancel and recreate them after upgrading if you want them to receive partition keys and run in parallel.
• Review custom archive code: Ensure any partner- or in-house-developed custom archive scenarios set a non-overlapping job criteria key (such as DataAreaId) so they don't silently default to single-file execution.
• Schedule around the 3/5 rule: Structure your archival batches across functional scenarios (such as combining three General Ledger jobs with two Inventory or Sales Order jobs) to maximize all five global concurrency slots.