You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: articles/search/search-sku-manage-costs.md
+21-32Lines changed: 21 additions & 32 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -19,6 +19,8 @@ As a first step, estimate your baseline costs by using the Azure pricing calcula
19
19
20
20
Azure provides built-in cost management that cuts across service boundaries to provide inclusive cost monitoring and the ability to set budgets and define alerts. The costs of running a search service will vary depending on capacity and which features you use. After you create your search service, optimize capacity so that you pay only for what you need.
21
21
22
+
<aname="billable-events"></a>
23
+
22
24
## Understand the billing model
23
25
24
26
Azure Cognitive Search runs on Azure infrastructure that accrues costs when you deploy new resources. It's important to understand that there could be other additional infrastructure costs that might accrue.
@@ -27,50 +29,37 @@ Azure Cognitive Search runs on Azure infrastructure that accrues costs when you
27
29
28
30
When you create or use Search resources, you're charged for the following meters:
29
31
30
-
+ You're charged an hourly rate based on the [pricing tier](search\search-sku-tier.md) of your search service, prorated to the hour.
32
+
+ You're charged an hourly rate based on the [pricing tier](search-sku-tier.md) of your search service, prorated to the hour.
31
33
32
-
+ The charge is applied to a *search unit* (SU) allocated to the service. Search units are the product of replicas and partitions (R x P = SU), which allows you to scale using flexible combinations of replicas and partitions, varying capacity depending on whether you need more query or indexing power.
34
+
+ The charge is applied per the number of search units (SU) allocated to the service. Search units are [units of capacity](search-capacity-planning.md) and are the product of replicas and partitions (R x P = SU) allocated to your service.
33
35
34
-
Billing is based on capacity and the costs of running premium features, such as [AI enrichment](cognitive-search-concept-intro.md), [Semantic search](semantic-search-overview.md), and [Private endpoints](service-create-private-endpoint.md).
36
+
Billing is based on capacity (SUs) and the costs of running premium features, such as [AI enrichment](cognitive-search-concept-intro.md), [Semantic search](semantic-search-overview.md), and [Private endpoints](service-create-private-endpoint.md). Meters associated with premium features are listed in the following table.
35
37
36
38
| Meter | Unit |
37
39
|-------|------|
38
-
| Indexer usage |1 unit for every 1000 API calls |
39
-
| Image extraction (AI enrichment) <sup>1</sup> | Per 1000 images. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing). |
40
-
| Custom Entity extraction (AI enrichment) <sup>2</sup> | Per 1000 text records. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing)|
41
-
| Custom Entity Lookup skill (AI enrichment) <sup>2</sup> | Per 1000 text records. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing)|
42
-
| Built-in skills (AI enrichment) <sup>2</sup> | Number of transactions, billed at the same rate as if you had performed the task by calling Cognitive Services directly. You can process 20 documents per indexer per day for free. Larger or more frequent workloads require a multi-resource Cognitive Services key. |
43
-
| Semantic Search <sup>2</sup> | Number of queries, billed at a progressive rate. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing). |
44
-
| Private Endpoints <sup>2</sup> | Billed as long as the endpoint exists, and billed for bandwidth. |
40
+
| Indexer usage |Per 1000 API calls |
41
+
| Image extraction (AI enrichment) <sup>1, 2</sup> | Per 1000 images. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing). |
42
+
| Custom Entity extraction (AI enrichment) <sup>1</sup> | Per 1000 text records. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing)|
43
+
| Custom Entity Lookup skill (AI enrichment) <sup>1</sup> | Per 1000 text records. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing)|
44
+
| Built-in skills (AI enrichment) <sup>1</sup> | Number of transactions, billed at the same rate as if you had performed the task by calling Cognitive Services directly. You can process 20 documents per indexer per day for free. Larger or more frequent workloads require a multi-resource Cognitive Services key. |
45
+
| Semantic Search <sup>1</sup> | Number of queries of "queryType=semantic", billed at a progressive rate. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing). |
46
+
| Private Endpoints <sup>1</sup> | Billed as long as the endpoint exists, and billed for bandwidth. |
45
47
46
-
<sup>1</sup> In an [indexer configuration](/rest/api/searchservice/create-indexer#indexer-parameters), **imageAction** is the parameter that triggers image extraction. If **imageAction** is set to "none" (the default), you won't be charged for image extraction. See the [pricing page](https://azure.microsoft.com/pricing/details/search/) page for image extraction charges.
48
+
<sup1</sup> Applies only if you use or enable the feature.
47
49
48
-
<sup>2</sup> Applies only if you enable the feature.
50
+
<sup>2</sup> In an [indexer configuration](/rest/api/searchservice/create-indexer#indexer-parameters), "imageAction" is the parameter that triggers image extraction. If "imageAction" is set to "none" (the default), you won't be charged for image extraction.
49
51
50
-
There is no meter on the number of queries, query responses, or documents ingested, although [service limits](search\search-limits-quotas-capacity.md) do apply at each tier.
52
+
There is no meter on the number of queries, query responses, or documents ingested, although [service limits](search-limits-quotas-capacity.md) do apply at each tier.
51
53
52
54
Data traffic might also incur networking costs. See the [Bandwidth pricing](https://azure.microsoft.com/pricing/details/bandwidth/).
53
55
54
-
Several premium features ([Knowledge store](knowledge-store-concept-intro.md), [Debug Sessions](cognitive-search-debug-session.md), [Enrichment cache (preview)](cognitive-search-incremental-indexing-conceptual.md)) have a dependency on Azure Storage. The meters for Azure Storage apply in this case, and the associated storage costs of using those features will be included in the Azure Storage bill.
56
+
Several premium features ([Knowledge store](knowledge-store-concept-intro.md), [Debug Sessions](cognitive-search-debug-session.md), [Enrichment cache (preview)](cognitive-search-incremental-indexing-conceptual.md)) have a dependency on Azure Storage. The meters for Azure Storage apply in this case, and the associated storage costs of using these features will be included in the Azure Storage bill.
55
57
56
58
[Customer-managed keys](search-security-manage-encryption-keys.md) provide double encryption of sensitive content. This feature requires a billable [Azure Key Vault](https://azure.microsoft.com/pricing/details/key-vault/)).
57
59
58
-
Skillsets can include [billable built-in skills](cognitive-search-predefined-skills.md), non-billable built-in utility skills, and custom skills.
59
-
60
-
+ Non-billable utility skills include Conditional, Shaper, Text Merge, Text Split. There is no billing impact when using them, no Cognitive Services key requirement, and no 20 document limit.
61
-
+ A custom skill is functionality you provide. The cost of using a custom skill depends entirely on whether custom code is calling other metered services. There is no Cognitive Services key requirement and no 20 document limit on custom skills.
62
-
63
-
<!--
64
-
60
+
Skillsets can include [billable built-in skills](cognitive-search-predefined-skills.md), non-billable built-in utility skills, and custom skills. Non-billable utility skills include Conditional, Shaper, Text Merge, Text Split. There is no billing impact when using them, no Cognitive Services key requirement, and no 20 document limit.
65
61
66
-
| Operation | Billing impact |
67
-
|-----------|----------------|
68
-
| Document cracking, text extraction | Free |
69
-
| Document cracking, image extraction | Billed according to the number of images extracted from your documents. |
70
-
| [Built-in skills](cognitive-search-predefined-skills.md) based on Cognitive Services | Billed at the same rate as if you had performed the task by using Cognitive Services directly. You can process 20 documents per indexer per day for free. Larger or more frequent workloads require a key. |
71
-
| [Built-in skills](cognitive-search-predefined-skills.md) that do not add enrichments | None. Non-billable utility skills include Conditional, Shaper, Text Merge, Text Split. There is no billing impact, no Cognitive Services key requirement, and no 20 document limit. |
72
-
| Custom skills | |
73
-
| [Custom Entity Lookup](cognitive-search-skill-custom-entity-lookup.md) | Metered by Azure Cognitive Search. See the [pricing page](https://azure.microsoft.com/pricing/details/search/#pricing) for details. | -->
62
+
A custom skill is functionality you provide. The cost of using a custom skill depends entirely on whether custom code is calling other metered services. There is no Cognitive Services key requirement and no 20 document limit on custom skills.
74
63
75
64
## Monitor costs
76
65
@@ -82,13 +71,13 @@ Follow these guidelines to minimize costs of an Azure Cognitive Search solution.
82
71
83
72
1. If possible, create all resources in the same region, or in as few regions as possible, to minimize or eliminate bandwidth charges.
84
73
85
-
1. Scale up for resource-intensive operations like indexing, and then readjust downwards for regular query workloads. If there are predictable patterns to your workloads, you might be able to synchronize scale up to coincide with the expected volume (you would need to write code to automate this).
74
+
1.[Scale up](search-capacity-planning.md) for resource-intensive operations like indexing, and then readjust downwards for regular query workloads. If there are predictable patterns to your workloads, you might be able to synchronize scale up to coincide with the expected volume (you would need to write code to automate this).
86
75
87
76
When estimating the cost of a search solution, keep in mind that pricing and capacity aren't linear (doubling capacity more than doubles the cost on the same tier). Also, at some point, switching up to a higher tier can give you better and faster performance at roughly the same price point. For more information and an example, see [Upgrade to a Standard S2 tier](search-performance-tips.md#tip-upgrade-to-a-standard-s2-tier).
88
77
89
-
1. Consider Azure Web App for your front-end application so that requests and responses stay within the data center boundary.
78
+
1. Consider [Azure Web App](../app-service/overview.md) for your front-end application so that requests and responses stay within the data center boundary.
90
79
91
-
1. If you're using AI enrichment, there is an extra charge for blob storage, but the cumulative cost goes down if you enable incremental enrichment.
80
+
1. If you're using [AI enrichment](cognitive-search-concept-intro.md), there is an extra charge for blob storage, but the cumulative cost goes down if you enable [enrichment caching](cognitive-search-incremental-indexing-conceptual.md).
0 commit comments