Enterprise Search Architectures

for SharePoint Server 2016 and SharePoint Server 2019

Overview Hardware requirements and scaling considerations
The requirements apply to each of the servers in the small, medium or large Enterprise Search topology. You can deploy search topologies for the enterprise on physical hardware or on virtual machines.
This model illustrates small, medium, and large-size farm architectures. The size of each farm
Search components
is based on the number of items that are crawled and included in the search index. Note: For evaluation use, you can place all search components on one server with 8GB RAM.
Architecture requirements can vary depending on the composition of the data that is crawled Index
(size of items and formats). The examples illustrate the type of search components needed
and how many of each. Use these examples as starting points for planning your own search
Index component Minimum hardware requirements for application servers Minimum hardware requirements for database servers
The index component is the logical representation of an index replica.
environments. For more information about search processes and how search components SEARCH COMPONENT HARD DISK RAM PROCESSOR HARDWARE COMPONENT REQUIREMENTS
interact, see Search Architectures for SharePoint® Server 2016 and SharePoint Server 2019 Index partitions 80GB regardless of the number of search components 32GB if the server hosts only an index 64-bit, 8 cores minimum. 64-bit, 4 cores for small deployments.

(https://go.microsoft.com/fwlink/p/?linkid=832554).
• You can divide the index into discrete portions, each holding a separate part of the index.
• An index partition is stored in a set of files on a disk. Index component
hosted on the server.
* component.
When hosting virtual machines on
Processor
64-bit, 8 cores for medium deployments.
500GB additional disk space, preferably a separate disk 32GB if the server hosts an index Windows Server 2008 R2 SP1, maximum 4 8 GB for small deployments.
• The search index is the aggregation of all index partitions. volume/partition. component and a query processing cores are possible. RAM

Search databases
component. 16 GB for medium deployments.

Index replicas 80GB regardless of the number of search components 8GB if the server hosts only one of 64-bit, 4 cores minimum, 8 cores
Hard Disk 80 GB for system drive.
*
Search admin DB
• Each index partition holds one or more index replicas that contain the same information.
hosted on the server.
* these search components. recommended.

• You have to provision one index component for each index replica. Analytics processing component
300GB additional disk space, preferably a separate disk 16GB if the server hosts two or more When hosting virtual machines on
volume/partition. This disk space is necessary for local of these search components. This Windows Server 2008 R2 SP1, maximum 4
Search administration database • To achieve fault tolerance and redundancy, create additional index replicas for each index processing of analytics data before it is written to the does not apply if the server hosts an cores are possible.
partition and distribute the index replicas over multiple application servers.
Stores search configuration data. Only one search administration database per Search service analytics reporting database. index component.
Scaling out for performance
application. Crawl component 80GB regardless of the number of search components

Query processing Content processing component

hosted on the server.
* Key performance metrics and scale actions

Crawl DB TO IMPROVE THIS METRIC TAKE THESE ACTIONS

Query processing component Query processing component Add more crawl databases and content processing components for result freshness.

Crawl database Analyzes and processes search queries and results. Search administration component Full crawl time and result freshness
Crawl databases and content processing components can be distributed among their respective servers.

Stores the crawl history and manages crawl operations. Each crawl database can have one or The Crawl Health Reports can be used to determine the cause of bottlenecks, if any.
more crawl components associated with it. Admin To improve query latency: Add more index replicas so that the query load is distributed more evenly among the
index replicas. This solution is better suited for small topologies.

Search administration component

Link DB Runs system processes that are essential to search. There can be more than one search * The server must have sufficient disk space for the base installation of the
Windows Server operating system and sufficient disk space for diagnostics such
as logging, debugging, creating memory dumps, and so on.
Time required for results to be returned To improve query latency and query throughput: Split the search index into more partitions to reduce the
number of items on each partition.
administration component per Search service application, but only one is active at any given
Link database time.
For production use, the server also needs additional free disk space for day-to-
day operations and for the page file. Follow the guidance on free disk space and
The Query Health Reports can be used to determine the cause of bottlenecks, if any.

Stores the information extracted by the content processing component and also stores click- page file size corresponding to your Windows Server installation. Availability of query functionality Deploy redundant (failover) query processing components on different application servers.

through information. Crawl Availability of content crawling, processing,

Use multiple crawl databases on redundant database servers.
and indexing functionality
Use multiple content processing components on redundant application servers.
Crawl component
Analytics DB Crawls content based on what is specified in the crawl databases.

Analytics reporting database Content processing

Stores the results of usage analytics.
Content processing component
Carries out various processes on the crawled items, such as document parsing and property Scaling out search components as number of items increase Redundancy and availability
mapping.
NUMBER OF INDEX COMPONENTS QUERY PROCESSING CONTENT ANALYTICS CRAWL CRAWL LINK ANALYSTICS SEARCH Redundant search components must be installed on separate failure domains. All of the
Analytics example topologies: Small, Medium, and Large have redundant configurations.
ITEMS AND PARTITIONS COMPONENTS PROCESSING PROCESSING COMPONENTS DATABASES DATABASE REPORTING ADMINISTRATION
Analytics processing component COMPONENTS COMPONENTS DATABASE COMPONENT Search database redundancy must be handled by the SQL server configuration. SQL
Carries out search analytics and usage analytics. Use 2 query processing Add 1 crawl Add 1 link database Add an analytics reporting Use 2 search administration
2008 R2 and SQL 2012 are supported.
Add 1 index partition per 20 components for redundancy. database per 20 per 60 million items database when analytics components for redundancy,
General Guidance
million items Above 80 million items, million items performance starts decreasing for all farm sizes
increase to 4. For redundant crawling and query processing, it is not necessary to have a redundant
2 components analytics processing component. However, if the non-redundant analytics processing
20 million 2 2 2 2 1 1 Variable 2
1 partition
component fails, the search results will not have optimal relevance until the failure is
8 components Variable recovered.
20-80 million 2 4 2 2 2 1 2
4 partitions
20 components
200 million 4 6 6 2 5 2 Variable 2
10 partition

Example search topologies

Small search farm (~20M Items) Medium search farm (~80M Items) Large search farm (~200M Items)
This farm is intended to provide the full functionality of SharePoint Server search with fault tolerance for up to 20 million items This farm is intended to provide the full functionality of SharePoint Server search with fault tolerance for up to 80 million items in This farm is intended to provide a dedicated search farm with fault tolerance for up to 200 million items in the search index.
in the search index. Two versions are illustrated.. the search index. To make this an all-purpose farm, add Web servers (not shown) and the additional application servers and
databases that are noted.
Dedicated search farm All purpose farm
Host A Host B Host C Host D Host A Host B

Application Servers
Application Servers

Host C Host D
This farm illustrates only the search components and can This farm incorporates the full functionality of SharePoint
serve as a dedicated search farm for one or more Server. Application Server Application Server Application Server Application Server Application Server Application Server Application Server Application Server

SharePoint farms. Dedicated search farms do not include Query processing Query processing Query processing Query processing

Web servers. Replica Index partition 0 Replica Replica Index partition 2 Replica Replica Index partition 0 Replica Replica Index partition 2 Replica

Application Server Application Server Application Server Application Server

Host B Application Server Application Server
Host B Application Server Application Server
Host A Host B Host 1 Host 2
Replica Index partition 1 Replica Replica Index partition 3 Replica Replica Index partition 1 Replica Replica Index partition 3 Replica
Application Server Application Server Web Server Web Server

Query processing Query processing

The search index is stored across replicas. Each replica for a given index partition
Replica Index partition 0 Replica contains the same data. The data within index replicas is stored in the file system on the
Web Server Web Server server. Each replica is a logical representation of an index component. Host E Host F Host G Host H Host I Host J
Application Server Application Server
Application Server Application Server Application Server Application Server Application Server Application Server
Crawl Crawl
Host C Host D Query processing Query processing
Admin Admin Application Server Application Server
All other Application Replica Index partition 4 Replica Replica Index partition 6 Replica Replica Index partition 8 Replica
Analytics Analytics Server Roles Application Server Application Server
All other Application All other Application
Content processing Content processing Server Roles Server Roles
Office Web Apps Server Office Web Apps Server * Application Server
Host B Application Server Application Server Application Server Application Server Application Server

Application Server Application Server For a multi-purpose farm, add VMs here to support other application server roles.
Replica Index partition 5 Replica Replica Index partition 7 Replica Replica Index partition 9 Replica
Analytics Analytics Combining all Application Server roles onto one VM requires Windows Server
Host C Host D 2012.
Content processing Content processing
All SharePoint Databases

Search admin DB
All SharePoint Databases

Redundant copies of all

* Office Web Apps Server VMs can share the same host as a SharePoint Web or
application server. Crawl
Admin
Crawl
Admin Host K Host L Host M Host N
Link DB databases using SQL
clustering, mirroring, or Application Server Application Server Application Server Application Server
Crawl DB SQL Server 2012
AlwaysOn Analytics Analytics Analytics Analytics
Analytics DB
Content processing Content processing Content processing Content processing
Host 3 Host 4 Host E Host F
Database Servers

Paired hosts for fault tolerance Application Server Application Server All SharePoint Databases All SharePoint Databases Application Server Application Server Application Server Application Server

Query processing Query processing Search admin DB Redundant copies of all Crawl Crawl Analytics Analytics

Replica Index partition 0 Replica Link DB databases using SQL Admin Admin Content processing Content processing
clustering, mirroring, or
Crawl DB SQL Server 2012
Application Server Application Server AlwaysOn
Crawl DB
Crawl Crawl
For a multi-purpose farm,
Analytics DB add other farm databases. Host O Host P Host Q Host R
Database Servers

Admin Admin All other SharePoint

Databases
Analytics Analytics All SharePoint Databases All SharePoint Databases All SharePoint Databases All SharePoint Databases

Content processing
Host 5
All SharePoint Databases
Content processing Search admin DB Redundant copies of all Link DB Redundant copies of all
Paired hosts for fault tolerance Analytics DB databases using SQL Crawl DB databases using SQL
clustering, mirroring, or SQL clustering, mirroring, or SQL
Analytics DB Server 2012 AlwaysOn Crawl DB Server 2012 AlwaysOn
Host 6 Crawl DB Crawl DB
Crawl DB
All SharePoint Databases

Search admin DB Redundant copies of all

Link DB databases using SQL
clustering, mirroring, or
Crawl DB SQL Server 2012 Paired hosts for fault tolerance Paired hosts for fault tolerance
AlwaysOn
Analytics DB
SharePoint DB
All other SharePoint
Databases

Paired hosts for fault tolerance

© 2016 Microsoft Corporation. All rights reserved. To send feedback about this documentation, please write to us at ITSPdocs@microsoft.com.
© 2012 Microsoft Corporation. All rights reserved. To send feedback about this documentation, please write to us at ITSPDocs@microsoft.com.