Azure quota requirements

Updated August 21, 2026

Art2link ESB installs into your own Azure subscription, so it can only create what your subscription is already allowed to create. Confirm the quota for your license tier, in the region you are deploying to, before you start the install.

Every tier provisions one App Service Plan and two Azure SQL databases, called SQL1 and SQL2 below. Grow and every Enterprise size add a fixed pair of Maintenance resources on top. Choosing Private AI adds three more. That is the whole list.

Step 1

Find your tier

What the install provisions

License tier	App Service Plan	SQL1	SQL2
Starter	B2	S0, 10 DTU	S0, 10 DTU
Grow	P0v4	S0, 10 DTU	S0, 10 DTU
Enterprise 2-core	P1mv4	S1, 20 DTU	S1, 20 DTU
Enterprise 4-core	P2mv4	HS_Gen5_2, 2 vCores	HS_Gen5_2, 2 vCores
Enterprise 8-core	P3mv4	HS_Gen5_2, 2 vCores	HS_Gen5_4, 4 vCores
Enterprise 16-core	P4mv4	HS_Gen5_4, 4 vCores	HS_Gen5_8, 8 vCores

Maintenance, on Grow and every Enterprise size

Grow and above also provision the Maintenance resources. These are fixed, identical on every tier from Grow up, and they do not scale with core count. Starter does not include Maintenance.

Resource	SKU
Maintenance App Service Plan	B1
Maintenance database	S0, 10 DTU

Grow and above need App Service quota in more than one family. The Maintenance plan is B1, which is Basic, while the tier plan is Premium v4. On Enterprise the tier plan is a memory-optimized Pmv4 SKU, so check that variant specifically. Clearing one family does not clear another.

Step 2

Add the Private AI resources, if you chose Private AI

The deployment wizard offers three AI options, and only one of them needs quota from you. Private AI deploys the model into your own subscription and adds the three resources below, in the same region. Managed AI calls the Accelerator hosted by Cerebrum City and provisions nothing here. No AI provisions nothing either. See AI deployment options for what each one means, and note that the choice cannot be changed after the install.

Private AI only

Resource	SKU or model
Azure AI Search	F0, Free
Foundry model deployment	gpt-5.4-mini, version 2026-03-17
Foundry model deployment	text-embedding-3-small

The F0 free search tier is capped per subscription and cannot be raised on request. If your free slot is already used, release it or provision a paid search tier.

Step 3

Check what you already have

You need Owner or Contributor on the subscription. Requests are free. Set Location to your deployment region on every check, because quota is granted per region.

Where to check each resource

Resource	Open	Then filter by
App Service Plans (tier plan, and B1 for Maintenance)	My quotas	Provider Microsoft.Web, then search Basic and Premium v4. Grow and above need both.
DTU databases (S0, S1, including Maintenance)	My quotas	Provider Microsoft.Sql, then search DTU
Hyperscale databases (HS_Gen5_*)	My quotas	Provider Microsoft.Sql, then search vCore. Counted separately from DTU.
Foundry models	Foundry quota page	Token per minute tab, then Request quota
Azure AI Search F0	Not raisable. See AI Search service limits.

If My quotas lets you raise it there, do that and you are done. If it does not, or the resource is not listed, use the forms in Step 4.

Step 4

Fill out the request

The forms ask for a region, a SKU and a new limit. These are the numbers to give them. The DTU figure is every Standard database on the tier added together, on the assumption they share one logical server.

What to ask for, by tier

License tier	App Service, Basic	App Service, Premium v4	SQL DTU, per server	SQL Hyperscale vCores
Starter	1 × B2	none	20 DTU	none
Grow	1 × B1	1 × P0v4	30 DTU	none
Enterprise 2-core	1 × B1	1 × P1mv4	50 DTU	none
Enterprise 4-core	1 × B1	1 × P2mv4	10 DTU	4 vCores
Enterprise 8-core	1 × B1	1 × P3mv4	10 DTU	6 vCores
Enterprise 16-core	1 × B1	1 × P4mv4	10 DTU	12 vCores

New limit is additive. Ask for your current limit plus the figure above, not the figure on its own, or you will be granted less than you had.

Why the Hyperscale tiers still need DTU. From Enterprise 4-core up, SQL1 and SQL2 are Hyperscale and consume vCore quota, but the Maintenance database stays S0. That is the 10 DTU in those rows. You need both allowances, not one.

Azure support request, for App Service and Azure SQL

Open a new support request and enter:

Field	Value
Issue type	Service and subscription limits (quotas)
Subscription	The subscription you are installing into
Quota type	App Service for the plan, SQL Database for the two databases. File them as two requests.
Location	Your deployment region
SKU or pricing tier	Your tier plan from Step 1, and B1 for the Maintenance plan on Grow and above
New quota or New limit	Current limit plus the figure from the table above

Azure SQL, which quota type to pick

If you need	Quota type	Enter
Hyperscale vCores, 4-core tier and up	vCores per subscription	New quota, current plus the vCore figure
DTU headroom on the logical server	DTUs	Server name and New limit in DTU
A region your subscription has never used	Region access	Expected Consumption in vCores
Zone redundancy in that region	Zone Redundant Access (Availability Zones)	Expected Consumption in vCores

Foundry model quota, Private AI only

Private AI only. On the Foundry quota page, Token per minute tab, use Request quota. For a large increase use the model quota increase form, which also asks for your subscription ID.

Field	Value
Model	gpt-5.4-mini, version 2026-03-17, and separately text-embedding-3-small
Region	Your deployment region, and the model must be offered there
Tokens per minute	Confirm the figure with Cerebrum City before filing

Small increases in a region with capacity are often approved automatically. Large increases, brand new subscriptions and constrained regions get routed to a person and can take a business day. Foundry quota changes need up to 15 minutes to propagate once approved.

Ready to install

Both App Service Plans, every database including Maintenance, and the model resources if you chose Private AI, all clear in your deployment region. See Azure Marketplace for the install itself.