| Limit
| Maximum value
| Limit type
| Notes
|
| Search service applications | 20 per farm | Supported | Multiple
Search service applications can be deployed on the same farm, because
you can assign search components and databases to separate servers. This
limit is lower than the limit for the total number of service
applications in a farm. |
| Crawl databases | 5 crawl databases per search service application | Boundary | The
crawl database stores the crawl data (time/status, etc.) about all
items that have been crawled. The supported limit is 5 crawl databases
per SharePoint Search service application. |
| Crawl components | 2 per search service application | Threshold | |
| Index components | 60 per Search service application | Supported | The
maximum number of index components that can be used is achieved by
multiplying the index partitions limit with the index replicas limit. |
| Index partitions | 20 per search service application | Supported | An
index partition holds a subset of the Search service application index.
Increasing the number of index partitions results in each partition
holding a smaller subset of the index, reducing the RAM and disk space
that is needed on the servers hosting the index components. |
| Index replicas | 3 per index partition | Supported | Each
index partition can have a set of replicas. If you increase the number
of index replicas this will have a positive effect on the query
performance and it provides better fault tolerance. However, if you add
too many replicas to your index partition, this can have a negative
effect on indexing. |
| Indexed items | 100 million per search service application; 10 million per index partition | Supported | Each
index partition contains a subset of the entire search index. If the
number of indexed items is high in relation to the amount of memory the
server has, this will affect the query response time negatively. |
| Crawl log entries | 100 million per search application | Supported | This is the number of individual log entries in the crawl log. It will follow the "Indexed items" limit. |
| Property databases | 10 per search service application;128 total | Threshold | The
property database stores the metadata for items in each index partition
associated with it. An index partition can only be associated with one
property store. The recommended limit is 10 property databases per
search service application. The boundary for index partitions is 128. |
| Link database | Two per Search service application | Threshold | Each link database can contain up to 50 million items. |
| Query processing components | 1 per server computer | Threshold | SharePoint only supports one query processing component per physical machine or virtual machine. |
| Content processing components | One per server computer | Supported | The topology supports scaling out the number of Content processing components.
Although
a specific physical host or virtual machine does support multiple
Content processing components, you achieve better usage of the CPU
capacity by using one Content processing component. The reason is that a
built-in mechanism maximizes CPU usage by adjusting the number of
feeding sessions in accordance with available CPU cores. Multiple
feeding sessions allow the Content processing component to process
incoming documents in parallel. This mechanism assumes a single Content
processing component per host.
If the number of physical cores on
the host equals N, then the Content processing component will have N*K
feeding sessions. K is a constant coefficient with the initial value 3. A
4-core server will have 12 feeding sessions, which means that the
Content processing component can process 12 documents in parallel. You
can change the value of K by setting the NumberOfCssFeedersPerCPUForRegularCrawl
property of the Search Service Application. SharePoint 2013 limits the
value of N upwards to 12, even if a server has more than 12 physical
cores. Therefore a 16-core server will have N*K = 12 * 3 = 36 feeding
sessions.
In the case that there still is idle CPU time,
consider increasing the K coefficient instead of adding an extra Content
processing component. If you increase the K coefficient, you must
ensure that the host has sufficient available memory. |
| Scope rules | 100 scope rules per scope; 600 total per search service application | Threshold | Exceeding this limit will reduce crawl freshness, and delay potential results from scoped queries. |
| Scopes | 200 site scopes and 200 shared scopes per search service application | Threshold | Exceeding
this limit may reduce crawl efficiency and, if the scopes are added to
the display group, affect end-user browser latency. Also, display of the
scopes in the search administration interface degrades as the number of
scopes passes the recommended limit. |
| Display groups | 25 per site | Threshold | Display
groups are used for a grouped display of scopes through the user
interface. Exceeding this limit starts degrading the scope experience in
the search administration interface. |
| Alerts | 100,000 per search application | Supported | This
is the limit for a Search service application with a mix of end user
queries (75%) and alert queries (25%). The limit for a Search service
application with only alert queries is 400,000 alerts. These limits are
based on a system with five queries per second (QPS). |
| Content sources | 50 per search service application | Threshold | The
recommended limit of 50 can be exceeded up to the boundary of 500 per
search service application. However, fewer start addresses should be
used, and the concurrent crawl limit must be followed. |
| Start addresses | 100 per content source | Supported | |
| Concurrent crawls | 20 per search application | Threshold | This is the number of crawls underway at the same time. Exceeding this number may cause the overall crawl rate to decrease. |
| Crawled properties | 500,000 per search application | Supported | The
contents and metadata of the items that you crawl are represented as
crawled properties. You can map these crawled properties to managed
properties. When the number of crawled properties exceeds this supported
limit, indexing speed will be reduced. |
| Crawl impact rule | no limit | Supported | |
| Crawl rules | no limit | Supported | |
| Managed properties | 50,000 per search service application | Supported | Search
uses managed propertied in queries. Crawled properties are mapped to
managed properties. When you exceed the supported limit for managed
properties, this reduces indexing speed. |
| Values per managed property | 100 | Supported | A
managed property can have multiple values of the same type. This is the
supported number of values per managed multi-valued managed property
per document. Exceeding this limit may have a negative effect on query
performance, index disk size and/or memory usage. |
| Indexed managed property size | 512 KB per searchable/queryable managed property | Threshold | This
is the default limit for the size of a searchable or queryable managed
property. If you increase this limit, you will enable indexing of more
data per managed property. Indexing more data per managed property uses
more disk space and increases the overall load on the system. You can
configure this limit by using Windows PowerShell cmdlets and the schema
object model to set the MP.MaxCharactersInPropertyStoreIndex attribute. |
| | | |
| Managed property mappings | 100 per managed property | Supported | Crawled properties can be mapped to managed properties. Exceeding this limit may decrease crawl speed and query performance. |
| Retrievable managed property size | 16 KB per managed property | Threshold | This
is the default maximum limit for the size of a retrievable managed
property. Increasing this limit will enable indexing of more data per
managed property. It will also enable retrieval of more data per managed
property with the search results. Indexing and retrieving more data per
managed property increases the overall load on the system and uses more
disk space. You can configure this limit per managed property by using
Windows PowerShell cmdlets and the schema object model to set the P.MaxCharactersInPropertyStoreForRetrieval attribute. |
| Sortable and refinable managed property size | 16 KB per managed property | Boundary | This
is the default maximum limit for the size of a sortable and refinable
managed property. Increasing this limit will enable indexing of more
data per managed property. This uses more disk space and increases the
overall load on the system. You can configure this limit per managed
property by using Windows PowerShell cmdlets and the schema object model
to set the MP.MaxCharactersInPropertyStoreForRetrieval attribute. |
| URL removals | 100 removals per operation | Supported | This is the maximum recommended number of URLs that should be removed from the system in one operation. |
| Authoritative pages | 1 top level and minimal second and third level pages per search service application | Threshold | The
recommended limit is one top-level authoritative page, and as few
second -and third-level pages as possible to achieve the desired
relevance.
The boundary is 200 per relevance level per search
application, but adding additional pages may not achieve the desired
relevance. Add the key site to the first relevance level. Add more key
sites at either second or third relevance levels, one at a time, and
evaluate relevance after each addition to ensure that the desired
relevance effect is achieved. |
| Keywords | 200 per site collection | Supported | The
recommended limit can be exceeded up to the maximum (ASP.NET-imposed)
limit of 5,000 per site collection given five Best Bets per keyword. If
you exceed this limit, display of keywords on the site administration
user interface will degrade. The ASP.NET-imposed limit can be modified
by editing the Web.Config and Client.config files
(MaxItemsInObjectGraph). |
| Metadata properties recognized | 10,000 per item crawled | Boundary | This
is the number of metadata properties that can be determined and
potentially mapped or used for queries when an item is crawled. |
| Analytics processing components | 6 per Search service application | Threshold | |
| Analytics reporting database | Four per Search service application | Threshold | Add
an analytics reporting database when the size of any of the deployed
analytics databases reaches 250 GB. This way repartitioning is as
balanced as possible. |
| Maximum eDiscovery KeywordQuery text length | 16 KB | Supported | The
Keyword Query Language is a query language for building search queries.
This is the default limit for the maximum text length of an eDiscovery
keyword query. |
| Maximum KeywordQuery text length | 4 KB | Supported | The
Keyword Query Language is a query language for building search queries.
This is the default limit for the maximum text length of a keyword
query. |
| Maximum length of eDiscovery KeywordQuery text at Search service application level | 20 KB | Boundary | The
Keyword Query Language is a query language for building search queries.
This is the maximum boundary for the text length of an eDiscovery
keyword query. This boundary is valid at the Search service application
level. |
| Maximum length of KeywordQuery text at Search service application level | 20 KB | Boundary | The
Keyword Query Language is a query language for building search queries.
This is the maximum boundary for the text length of a keyword query.
This boundary is valid at the Search service application level. |
| Maximum size of documents pulled down by crawler | 64 MB (3 MB for Excel documents) | Boundary | |
| Navigable results from search | 100,000 per query request per Search service application | Supported | This is the limit for how many hits that a query requests. Increasing this limit will affect the query performance negatively. |
| Number of entries in a custom entity extraction dictionary | 1 million | Supported | This is the tested limit. |
| Number of entries in a custom search dictionary | 5,000 terms per tenant | Boundary | This
limits the number of terms allowed for inclusions and exclusions
dictionaries for query spelling correction and company extraction. |
| Number of entries in a thesaurus | 1 million | Supported | This is the tested limit. |
| Ranking models | 1,000 per tenant | Boundary | Approaching this limit can have negative effect on the overall system performance. |
| Results removal | No limit | Supported | |
| Term size | 300 characters | Boundary | Terms
that are longer than this boundary will be split into two or more terms
where no term has more than 300 characters. For example, a
612-character term will be split into two 300-character terms and one
12-character term. Only the first 1000 characters of a term are
considered for splitting, any remaining characters are ignored. |
| Unique terms in the index | 2^31 (>2 billion terms) | Boundary | This is the maximum boundary for the number of unique terms that can exist in the index of a Search service application. |
| Unique contexts used for ranking | 15 unique contexts per rank model | Boundary | This is the maximum boundary for the number of unique contexts per rank model. |
| User defined full text indexes | 10 | Boundary | This is the maximum boundary for the number of full text indexes. |