1. Infrastructure Layer (The "Nonstop Cloud")

The foundation of the platform relies on a unique Multi-Instance architecture rather than the standard Multi-Tenant model found in most SaaS environments.

Multi-Instance Architecture

Unlike multi-tenant architectures where all customers share one massive database, ServiceNow assigns unique resources to every customer.

• Data Isolation: Your data is stored in its own database, not commingled with other customers.
• Performance Isolation: Heavy processing or queries by another customer will not impact your instance's performance.

Hardware & High Availability (AHA)

ServiceNow runs on bare-metal servers in their own data centers, with increasing usage of public cloud providers (Azure/GCP) for specific regions.

• Load Balancing: F5 Load Balancers distribute traffic across multiple application nodes to manage throughput.
• Real-Time Replication: Data is replicated in near real-time between a primary data center and a secondary (failover) data center within the same geography.
• Surgical Failover: The ability to move a specific instance to a backup data center without migrating other customers.

2. Database Layer

ServiceNow abstracts the database layer, allowing developers to work with objects while the platform manages physical storage and query optimization.

Primary Database (MariaDB)

The core database engine for transactional workloads (OLTP) is MariaDB (a fork of MySQL).

• It utilizes a relational model.
• Uses TPP (Table Per Partition) and TPC (Table Per Class) models to manage how extended tables (like cmdb_ci_server) are physically stored on disk.

Analytics Database (RaptorDB)

Introduced to handle high-volume analytics and reporting, RaptorDB utilizes PostgreSQL-based technologies. It supports Hybrid Transactional/Analytical Processing (HTAP), allowing queries to run significantly faster than standard MariaDB, reducing the need for external data lakes.

3. Application Runtime (The "Back-end")

This layer executes the server-side logic and core platform kernel.

• Java (J2EE) & Tomcat: The application nodes run on Apache Tomcat. The core platform "kernel" is written in highly optimized Java, which is inaccessible to customer developers.
• Rhino (JavaScript Engine): Server-side JavaScript (Business Rules, Script Includes) executes on the Mozilla Rhino engine embedded within the Java application.
• ECMAScript Support: Modern instances support ES12 (ECMAScript 2021), enabling modern syntax such as let, const, arrow functions, and template literals.
• Glide Layer: The bridge between Java and JavaScript. Objects like GlideRecord allow JavaScript to call underlying Java methods for database interaction.

4. Application Logic & Automation

The layer where the majority of development and business logic implementation occurs.

• Flow Designer: The modern replacement for the legacy Workflow engine. It runs on asynchronous Java threads to process logic efficiently.
• Integration Hub: Built on Flow Designer, this uses Java-based "Spokes" to handle protocols like REST, SOAP, JDBC, PowerShell, and SSH.
• Scope Isolation: Applications can be built in unique "Scopes," providing a runtime protection layer that prevents cross-application data modification without explicit privileges.

5. Frontend / User Experience (The "Now Experience")

The user interface has transitioned from proprietary scripts to standard web technologies under the Polaris (Next Experience) architecture.

• Seismic Framework: A custom JavaScript framework built on standard Web Components, using Shadow DOM for encapsulation.
• UI Builder (UIB): A low-code WYSIWYG tool used to generate the JSON configuration for Seismic pages.
• GraphQL: The primary communication protocol for the modern UI, allowing efficient data fetching compared to the legacy XML/AJAX methods.
• Legacy Fallback: Classic UI forms and lists still rely on Jelly Scripting and AngularJS (1.x), though these are being phased out.

6. Intelligence Layer (GenAI)

Artificial Intelligence is deeply integrated into the platform architecture.

• Now LLM: A domain-specific Large Language Model trained specifically on ServiceNow workflows, code, and documentation.
• AI Controller: A mediation layer connecting the instance securely to external LLMs (such as OpenAI or Google Gemini) for tasks like summarization and code generation.

Summary of the Technology Stack