These system requirements contain detailed information about the software and hardware you use to host Relativity in your environment and in the cloud. These requirements also provide various recommendations for configuring a new deployment of Relativity, as well as scaling your environment as the number of users and the amount of data continue to grow.
This page contains the following sections:
- Infrastructure overview
- Scalability
- Required configurations for new deployments
- Infrastructure configuration
- Software requirements
- Licensing Microsoft products
See these related pages:
Infrastructure overview
Relativity is designed with a scalable infrastructure that you can tailor to the requirements of your environment. It is developed on the .NET framework with a Microsoft SQL Server back-end. As illustrated in the following diagram, all areas of the platform are scalable providing support for any hardware vendor, hypervisor, and storage protocol.
The Web Server is the gateway for all users to access Relativity. It authenticates the user with the system, contains APIs for searching and third-party applications, transfers documents to the end user in the Relativity Viewer, and is responsible for communications during imports and exports in workspaces. There are different mechanisms for authentication into the system including forms, active directory, two-factor, SAML 2.0, and OpenID Connect.. User sessions can be load balanced with the included Relativity User Load Balancer or via available hardware load balancing solutions.
Agents in Relativity are responsible for running all background processing tasks. When a user submits a job, such as a Production or OCR job, the associated agent(s) will pick up the job and complete the work. The agents run under a Windows Service and often require various levels of CPU, RAM and I/O, depending on the job type. The agents can be scaled vertically and horizontally to accommodate organizational needs.
Agent server (conversion)
In Relativity 9.6 and above, viewer conversion jobs are handled by the Conversion agents. Any Relativity agent server designated as a conversion agent server should only have one conversion agent deployed. Conversion jobs are multi-threaded and one conversion agent may utilize all available processor cores on a server.
For more information, see Required configurations for new deployments.
Agent server (dtSearch)
dtSearch queries are multi-threaded and spawn as many threads as there are sub-indexes or cores — whichever number is lowest will be the constraint. One dtSearch search agent may be able to utilize all available processor cores on a server. Therefore, each Relativity agent server that is designated to be a dtSearch search agent server should only have one dtSearch search agent and nothing else.
Agent server (PDF)
PDF worker agents are multi-threaded. The number of threads are determined by the number of processors on the VM divided by 2 (e.g. an 8 core VM results in a PDF worker with 4 threads).
For optimal performance Relativity recommends that dedicated agent VMs are created with a maximum of 2 PDF worker agents per machine. This will allow the worker agents to maximize resource usage without performance degradation. If PDF worker agents are added to a shared agent server, only one should be added to limit the impact to other agents on the server.
SQL Server (workspaces)
This SQL Server is where the structured text and metadata resides for the documents. Each Relativity workspace is represented by its own SQL Server database. Environments may have one or more SQL Servers. In addition to workspace databases there are Relativity system databases present on each server that contain tables for system configurations, agent job queues, users/groups, etc.
SQL Server (Invariant/Worker Manager server)
Processing has individual store databases that correspond to each Relativity workspace database with processing enabled. Total memory and processor requirements for this role are not as demanding as the SQL Servers that house workspace databases. This server is also used for native imaging and save as PDF request management.
Worker
The ‘Worker’ role is responsible for handling enhanced native imaging and processing jobs. Relativity has placed a hard cap on the amount of threads that each Worker server is allowed to spawn, the hard cap is 16 threads. Each processor core and 2GB RAM will create two threads. Therefore, it is suggested that 8 logical cores and 16GB RAM be allocated to each worker server to get the most throughput.
Note: Refer to the Processing page for a breakdown of performance metrics.
Secret Store
The Secret Store is a required component that provides secure, auditable storage for Relativity secrets. A secret could be user credentials, a database connect string, an instance setting that contains confidential information such as your SMTP credentials, or a TLS certificate. All confidential information is stored securely in the Secret Store database that can be accessed only from authenticated servers.
For more information, see Relativity Secret Store.
Message broker
Relativity utilizes RabbitMQ as a message delivery service that communicates information about agent jobs to different application components. This infrastructure feature supports this communication by routing messages between application components. For example, Relativity uses RabbitMQ for submitting conversion jobs to agents and returning converted documents.
For more information, see Relativity Service Bus documentation.
Analytics
The analytics server is responsible for building and storing the conceptual indexes in the environment. Once an index is built, the server is also used to run the conceptual features such as categorization and clustering. In addition to conceptual indexing, structured analytics sets are run on this server for textual analysis features such as email threading or language identification. The indexes and structured analytics sets are stored on disk in a configurable location.
File server
This server may not be required depending on the available storage. Relativity doesn't install any software on a file server for Relativity. Relativity just needs to know where the files (Natives/Images) live and the web servers need to be able to access those locations. The same applies to dtSearch, Analytics index, and viewer cache location(s).
Data Grid master node
This is the server within a cluster that manages changes across the entire cluster.
Data Grid client node
This is the server that serves as the gateway through which data enters a cluster. When there is more than one in an environment, these can be thought of as load balancers which service requests for data.
Data Grid data node
This is the server that stores data within a cluster.
Scalability
Scalability
You can scale Relativity installations to handle the performance, storage, and other environmental factors necessary to support the addition of new users, continual growth of data, and increased demands for searching capabilities.
Tier level definitions
We have identified tier levels that support varying numbers of users and sizes of active data. You can use these tier level definitions to determine the cores, RAM, and other equipment required to support the rapid growth of your Relativity installation. Key terms used in the following table include:
- Enabled User Accounts – amount of enabled Relativity User accounts.
- Simultaneous Users - average amount of simultaneous users logged into Relativity.
- Active SQL Data (TB) - total amount of disk space consumed by SQL databases (mdf) and full text (ndf) indexes.
- Active Record Count (MM) – total amount of records (documents) included across all active Relativity workspaces.
- Active File Size (TB) - total amount of disk space consumed by native and image files.
This table identifies the combination of users, data, and file sizes associated with each tier.
|
Tier 1 - Entry Level Environment |
Tier 2 - Mid Level Environment |
Tier 3 - Large Scale Environment |
|
|---|---|---|---|
| Enabled User Accounts | < 300 | 300 - 1000 | 1000+ |
| Simultaneous Users
|
< 100 | 100 - 500 | 500+ |
| Active SQL Data (TB)
|
< 1 | 1 - 10 | 10+ |
| Active Record Count (MM) | < 20 | 20 - 100 | 100+ |
| Active File Size (TB)
|
< 5 | 5 - 30 | 30+ |
The equipment used to support environments at each tier is described in the following table.
|
Tier 1 - Entry Level Environment |
Tier 2 - Mid Level Environment |
Tier 3 - Large Scale Environment |
|
|---|---|---|---|
| Total Cores for Non-SQL | < 48 | 48 - 192 | 192+ |
| Total Memory (GB) for Non-SQL | < 96 | 96 - 384 | 384+ |
| Total Cores for SQL Server | < 16 | 16 - 96 | 96+ |
| Total Memory (GB) for SQL Server | < 128 | 128 - 1024 | 1024+ |
| Total SQL Storage I/O (Gbps) | 4 - 8 | 8+ | 16+ |
| SQL Tempdb Storage | SSD or flash | SSD or flash | SSD or flash |
Required configurations for new deployments
Required configurations for new deployments
Contact Client Services for assistance with designing your Relativity infrastructure.
- Notes:
- The following Tier 1 example environments provide information for different user and data counts. Most new deployments adhere to one of these Tier 1 examples.
- Refer to the Processing page to determine if your processing needs can be achieved with your selected number of named users. For example, you have 30 named users but process a heavy amount of data daily. For best performance, add additional workers to your current tier or move up a tier.
Tier 1 - Hardware requirements
25-50 named users
We support the installation of all Relativity components on a single device for 25-50 named user agreements. We also require that a hypervisor is installed to this device so each Relativity role has its own virtual machine.
The following table provides virtual machine specifications for the single server setup.
| Tier 1 (25-50 named users) - Single Server Deployment |
Quantity |
Memory (GB) |
CPU |
|---|---|---|---|
| Web | 2 | 16 | 4 |
| Agent (core) | 2 | 4 | 4 |
| Agent (dtSearch) | 1 | 4 | 4 |
| Agent (conversion) | 1 | 8 | 4 |
| Agent (PDF)* | 1 | 16 | 8 |
| Analytics | 1 | 32 | 4 |
| Worker | 1 | 16 | 8 |
| Secret Store | 1 | 4 | 4 |
| SQL (workspace databases) | 1 | 64 | 8 |
| SQL (Invariant/Worker Manager server) | 1 | 16 | 4 |
| Data Grid master/data/client node | 1 | 16 | 4 |
| Message broker server | 1 | 4 | 4 |
Note: * If users have a dedicated Invariant worker for Save As PDF jobs they can either decommission it, or convert it to an agent server following the recommendation of 2 PDF Worker agents per VM.
While Relativity supports the use of virtual machines, differing configurations are required depending on the tier level of your environment. The number of SQL Servers is influenced by High Availability and Disaster Recovery requirements as described in Infrastructure configuration. You can distribute Relativity workspace databases across multiple SQL instances.
This table lists the recommendations for environments at Tier 1.
| Tier 1 (100+ named users) - Entry Level Environment |
Quantity |
Memory (GB) |
CPU |
|---|---|---|---|
| Web | 2 | 16 | 8 |
| Agent (core) | 2 | 4 | 4 |
| Agent (dtSearch) | 1 | 4 | 4 |
| Agent (conversion) | 1 | 16 | 8 |
| Agent (PDF)* | 1 | 16 | 8 |
| Analytics | 1 | 32 | 8 |
| Worker | 2 | 16 | 8 |
| Secret Store | 1 | 4 | 4 |
| Message broker server | 1 | 8 | 8 |
| SQL (workspace databases) | 1 | 64 | 8 |
| SQL (Invariant/Worker Manager server) | 1 | 16 |
4 |
| Data Grid Mast Node/Data Node/Client Node | 3 | 16 | 4 |
Note: * If users have a dedicated Invariant worker for Save As PDF jobs they can either decommission it, or convert it to an agent server following the recommendation of 2 PDF Worker agents per VM.
Tier 2 - Hardware requirements
300 or more named users
For Tier 2 environments, additional virtual machines are required as well as increased RAM and CPUs as illustrated in the following table. Additionally, it is suggested that SQL (Workspace Databases) instances are not virtualized when supporting larger datasets.
| Tier 2 (300+ named users) - Mid Level Environment |
Quantity |
Memory (GB) |
CPU |
|---|---|---|---|
| Web | 2 | 32 | 8 |
| Agent (core) | 3 | 16 | 8 |
| Agent (dtSearch) | 2 | 16 | 8 |
| Agent (conversion) | 2 | 16 | 8 |
| Agent (PDF)* | 2 | 16 | 8 |
| Analytics (structured analytics) | 1 | 32 | 8 |
| Analytics (Analytics indexing) | 1 | 32 | 4 |
| Worker (processing, imaging) | 2 | 16 | 8 |
| Secret Store | 1 | 4 | 4 |
| Message broker server | 1 | 16 | 16 |
| SQL (workspace databases) | 2 | 256 | 16 |
| SQL (Invariant/Worker Manager server) | 1 | 32 | 4 |
| Data Grid master node | 3 | 32 | 8 |
Note: * If users have a dedicated Invariant worker for Save As PDF jobs they can either decommission it, or convert it to an agent server following the recommendation of 2 PDF Worker agents per VM.
Please take the following into consideration for any sized environment:
Note: For the File (Document) role, the type of storage system used will determine if you need to install the Windows operating system.
| File (Document) |
|
| SMTP (Notification) | Relativity requires an active SMTP server on your network. It interfaces with this server to send notifications and monthly billing statistics. The hardware requirements for this role are minimal. You can leverage an existing SMTP server in the network or merge this server with the agent server role. |
Storage
For each type of data, the amount of recommended space depends on the number of records imported, as well as the type and length of the expected reviews. Each server or VM needs space for the OS, page file and Relativity installation files. For the Processing SQL Server, all the same SQL data is required with the exception of SQL Full Text Indexes. Throughput, especially when multiple SQL Servers or Data Grid Data Nodes are virtualized on a single host, should be put through a regiment of rigorous random and sequential read/write IO testing before installation of Relativity is completed.
Recommended space by data type:
| Data Type | 25-50 Named Users | 100+ Named Users | Disk I/O |
|---|---|---|---|
| SQL databases | 500 GB | 750 GB | High |
| The databases can live across multiple storage volumes and SQL instances. | |||
| SQL full text indexes | 150 GB | 250 GB | Moderate |
| Index size depends on the number of fields and records indexed. | |||
| SQL database logs | 150 GB | 250 GB | High |
| Regular transaction log backups keep these values small and provide point in time recovery. | |||
| SQL Tempdb | 80 GB | 80 GB | High |
| We recommend eight 10GB Tempdb data files for new deployments. SSDs recommended. | |||
| SQL backups | 500 GB | 1000 GB | Low-High |
| We recommend having a backup strategy. This volume is not required. Larger data sizes may require higher I/O throughput. | |||
| dtSearch indexes | 150 GB | 250 GB | Moderate |
| The dtSearch index share is typically stored in the same location as the files. | |||
| Analytics indexes | 150 GB | 250 GB | High |
| The Analytics index volume is mounted to the Analytics virtual machine with speed and connectivity similar to that of SQL Server. | |||
| Files (natives/images) | 1500 GB | 3000 GB | Low-High |
| The files may not require a Windows installation depending on the storage. Multiple Processing Workers online will require more file storage I/O. | |||
| Viewer cache | 500GB | 1000GB | High |
| Temporarily store natives, images, productions, and other file types the viewer uses. It is recommended that the cache be stored on tier-one storage (SSDs) in environments with hundreds of concurrent users. Recommended 1TB viewer cache space available for every 100 concurrent users. | |||
| Agent (conversion) | 250 GB | 250 GB | High |
| This is the Windows temp directory used during document conversion. | |||
| Worker (native imaging/processing) |
250 GB | 250 GB | High |
| This is the Windows temp directory used by native applications during imaging and processing. This temp location never exceeds 250GB for each Worker server. | |||
| Data Grid data node | 1 TB | 1-10 TB | High |
Infrastructure configuration
Relativity supports the following technologies as part of its infrastructure configuration:
- Virtualization
All Aspects of Relativity can be virtualized. Some roles are more RAM and CPU intensive as others. You have to make sure when virtualizing Relativity that you do not overcomitt the RAM and CPU on a virtual machine to host ratio.
- High Availability (HA)
Relativity supports SQL Server Failover and File Server Clustering along with available hypervisor solutions. - Disaster Recovery (DR)
Relativity supports Microsoft SQL Server mirroring, log shipping, and SAN replication technologies. These approaches typically require manual failover and increased downtime. - Web Server Load Balancing
Relativity supports only single affinity in Windows Network Load Balancing (NLB). The Relativity User Load Balancer (RULB) provides the ability to distribute the user load evenly. - Perimeter Networking (DMZ)
Relativity requires certain ports to remain open for proper server communication. For more details, download the Relativity Infrastructure Ports Diagram from the Relativity Community. Note that you must have a valid username and password to download this content.
Guides for infrastructure management
Review the following guides to become familiar with best practices for managing the Relativity infrastructure:
-
-
-
- Relativity Server 2021 Deployment Workbook
Software requirements
Relativity has specific software requirements for servers or virtual machines, user workstations, and the Relativity Desktop Client. The requirements for servers differ by the role assigned to them in your system configuration.
Note: Make sure that you install the latest service packs and updates for your Windows Operating system and the latest service packs and cumulative updates for your SQL Server. However, compatibility for higher .NET versions is not guaranteed and we do not recommend installing higher .NET versions than what is listed as required by your Relativity version.
System (servers or virtual machines)
Note: Relativity is compatible with local settings Only for webservice servers.
The general software requirements for servers and virtual machines include Microsoft Windows Server and .NET technologies. Microsoft Office and other applications are required for worker servers.
The following table provides software requirements by server role.
| Server Role |
Software Requirements |
||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Web |
|
||||||||||||||||||||||||||||||||||||
| Agent |
|
||||||||||||||||||||||||||||||||||||
| Analytics |
|
||||||||||||||||||||||||||||||||||||
| Secret Store |
|
||||||||||||||||||||||||||||||||||||
| Message broker |
|
||||||||||||||||||||||||||||||||||||
| SQL |
|
||||||||||||||||||||||||||||||||||||
| Worker |
* Microsoft Project and Visio are not required to install and use Relativity Processing. These components are only required if you intend to process Project and Visio files, specifically. |
Workstations (end-user PCs)
In Relativity, end users perform their reviews on workstations. Each workstation should be configured with a browser in which to use the Relativity web application, an operating system on which to run the Relativity Desktop Client, and the currently supported version of .NET.
|
Supported browsers for Relativity Web application |
Supported operating systems - Relativity Desktop Client |
Supported operating systems - servers and Relativity Desktop Client |
Supported .NET version |
|---|---|---|---|
Note: As of August 31, 2017, we no longer support Internet Explore (IE) 10. Please upgrade to a compatible version of IE 11. |
|
|
|
Note: SQL Server 2019 requires Windows Server 2016 or 2019. It is not compatible with Windows Server 2012.
Supported .NET version
Relativity Desktop Client
The Relativity Desktop Client (RDC) is a utility used for importing and exporting documents, images, natives, and productions. This utility requires the following software:
The Relativity Desktop Client requires Microsoft .NET 4.6.2 or above and Visual C++ 2015 Redistributable Update 3 RC.
Your operating system determines whether you need to download the 64-bit or 32-bit version of these applications:
- If you're running a 32-bit machine, you must install the RDC 32-bit and the Visual C++ 2015 Redistributable Update 3 RC. For more information, see Microsoft Visual C++ 2015 Redistributable Update 3 RC.
- If you're running a 64-bit machine, you want to install the RDC 64-bit and the Visual C++ 2015 Redistributable Update 3 RC You may notice a significant improvement in the speed of the RDC with the 64-bit version. However, a 64-bit machine can have both the x86 and x64 redistributables installed at the same time, and it can run the 32-bit or 64-bit version of the RDC.
Licensing Microsoft products
Relativity requires Microsoft Windows and Microsoft SQL Server, both of which you need to license through Microsoft or one of their resellers. If using Processing or Native Imaging, you also need to license Microsoft Office, Visio, and Project through Microsoft or one of their resellers.
If Relativity is hosted for external customers, you may need to license Microsoft products through Microsoft’s SPLA (Service Provider License Agreement). You can find more information about Microsoft’s SPLA program on Microsoft’s Hosting site.
Note: We recommend contacting Microsoft, or one of their resellers, for guidance on the licensing options available.
