Reference SOC architecture: Logs, SIEM, SOAR, EDR/XDR, and the response layer
A reference SOC architecture can be organized into several layers: telemetry sources, collection and normalization, detection and analytics, SIEM and XDR, investigation and case management, orchestration with SOAR, response, and reporting. These layers are not isolated. A SOC becomes operationally effective when security data from endpoints, networks, identities, cloud environments, and applications is transformed into signals, investigated in context, escalated through defined workflows, and converted into controlled response actions.
SOC architecture is often illustrated as a collection of security products: a SIEM at the center, EDR on one side, SOAR above it, and threat intelligence connected somewhere nearby. That view can be useful for showing a technology stack, but it can also create the impression that building a SOC is mainly about purchasing and connecting tools.
In practice, a SOC is better understood as a data and decision operating model. Telemetry must be generated by the right systems, collected with enough context, processed through detection logic, turned into alerts, investigated by analysts, and then connected to escalation and response. If one of these stages is missing, the technology may still function while the overall security operations capability remains incomplete.
The reference model below is therefore vendor-neutral. It is designed to explain how telemetry, collection, SIEM, EDR/XDR, SOAR, case management, analysts, response, and governance fit together. It should not be treated as an official taxonomy from NIST or any specific security vendor.
What are the layers in a reference SOC architecture?
At a high level, a SOC architecture can be divided into seven functional layers:
Telemetry Sources
Collection & Normalization
Detection & Analytics
Investigation & Case Management
Orchestration & Automation
Response
Reporting & Governance

The operating flow can be simplified as:
Security Data → Collection → Detection → Investigation → Orchestration → Response → Governance
The key point is that these layers do not necessarily require seven different products. Modern security platforms often cover multiple functions at once. A SIEM may also provide investigation workflows, automation, or case management. An XDR platform may detect, investigate, and respond directly without requiring every event to pass through a SIEM first.
That is why SOC design should begin with required capabilities and use cases, not with the number of tools in the architecture.
Organizations that need a broader introduction to the operating role of a SOC can also review SOC (Security Operations Center): the digital fortress for Vietnamese enterprises in the 2026 cyber landscape
Figure: Reference SOC architecture showing the layered flow from telemetry sources and data collection to SIEM/EDR/XDR, investigation, SOAR, response, and governance.
Layer 1 - Telemetry Sources: What Can the SOC Actually See?
A SOC cannot detect activity that is never recorded or never reaches the monitoring environment.
Security telemetry may come from endpoints, identity systems, network devices, cloud platforms, email systems, applications, databases, SaaS environments, and security controls.
Each source provides a different view of activity.
Endpoint telemetry may show process execution, file changes, command activity, or EDR detections. Identity telemetry may reveal authentication events, privilege changes, session activity, or abnormal sign-in behavior. Network telemetry can provide visibility into connections, traffic patterns, domains, or protocol activity. Cloud telemetry may include audit events, configuration changes, API activity, and workload access.
The objective, however, should not be to collect every available log.
Collecting large volumes of telemetry without clear use cases can increase storage and processing costs without materially improving detection.
A more effective design approach starts with the question:
What behavior does the SOC need to detect or investigate?
From there, the organization can determine which telemetry sources are necessary.
Layer 2 - Collection and Normalization: Turning Raw Data Into Usable Security Context
Security data rarely arrives in one consistent format. A firewall may use one schema, an identity platform another, and a legacy application may only generate flat text logs. Cloud services often rely on APIs, while endpoint products may send telemetry through agents.
The collection layer is responsible for bringing these sources into the monitoring architecture through mechanisms such as agents, syslog, APIs, collectors, or event streams.
The next challenge is making the data usable. Parsing and normalization allow events from different systems to be interpreted consistently. Enrichment can add context such as asset ownership, business criticality, geolocation, identity information, or threat intelligence.
For example, a raw event stating only: Login failed - has limited value. The same event becomes much more useful when it includes: user + source IP + device + location + timestamp + authentication method + asset context
The quality of downstream detection is heavily dependent on the quality of this upstream data. In other words: Poor telemetry produces poor detection.
Layer 3 - detection and analytics: where data becomes a security signal
Once telemetry has been collected, the SOC needs a way to determine which events deserve attention. Detection may be based on:
correlation rules;
behavioral analytics;
anomaly detection;
threat intelligence matching;
detection engineering;
known indicators;
custom use cases.
A single failed login may be harmless. Hundreds of failed logins across multiple accounts, followed by a successful authentication and unusual access to sensitive resources, may be far more meaningful.
This is where correlation becomes important. The SOC does not only need to know that an event occurred. It needs to understand how multiple events relate to each other.
Detection rules also require ongoing maintenance. Environments change, applications are updated, attack techniques evolve, and false positives accumulate. Detection engineering therefore becomes a continuous operational capability rather than a one-time configuration task.
Where does SIEM fit in SOC architecture?
SIEM typically sits near the center of the data collection, analytics, and investigation workflow. A SIEM may ingest data from multiple sources, normalize events, perform correlation, generate alerts, support search and threat hunting, and provide reporting.
Modern SIEM platforms can also extend into automation, investigation, case management, or response. That overlap is important because SOC architecture is no longer built from completely separate technology categories.
However, a SIEM should not be mistaken for the SOC itself. The SOC still requires analysts, workflows, escalation paths, response authority, and governance.
A useful way to think about SIEM is as a technology layer that may span several capabilities, while the SOC is the broader operating system around those technologies.
Where do EDR and XDR fit?
EDR focuses primarily on endpoint activity.
It can provide telemetry about processes, files, command execution, connections, and suspicious behaviors on workstations and servers. Depending on the platform and permissions, EDR may also perform response actions such as isolating a host, terminating a process, or quarantining a file.
XDR expands this concept across multiple security domains. Depending on the platform, XDR may correlate endpoint, identity, email, cloud, and network signals into broader incidents.
This means EDR and XDR can play several roles at once. They may act as:
telemetry sources;
detection platforms;
investigation tools;
response mechanisms.
That is also why real SOC data flows are rarely perfectly linear. An XDR platform may create an incident directly and send it to analysts while also forwarding telemetry to a SIEM for broader correlation.
Organizations should therefore avoid simplistic assumptions such as: XDR always replaces SIEM or SIEM always replaces XDR. The two may overlap, but their roles depend on the operating environment and use cases.
Where does SOAR fit and what does it actually do?
SOAR stands for Security Orchestration, Automation and Response. Within a SOC architecture, SOAR usually sits between detection, investigation, and response.
Its primary role is to connect workflows and security systems so analysts do not have to perform every step manually. A simplified workflow might look like:
SIEM Alert → Enrichment → Reputation Check → Endpoint Query → Case Creation → Analyst Review → Approval → Endpoint Isolation
Without orchestration, an analyst may need to open multiple consoles, copy IP addresses or hostnames between tools, search threat intelligence manually, query endpoint systems, create a ticket, and notify stakeholders one step at a time.
SOAR allows parts of this workflow to be standardized in playbooks. However, not every response action should be fully automated.
Automating IP enrichment or asset lookup may be relatively low risk. Automatically disabling a privileged account or isolating a production server can have significant business impact and may require analyst or business approval.
SOAR should therefore be understood as a way to orchestrate and automate selected workflows, not as a mechanism for allowing machines to handle every incident without human control.
How is case management different from SIEM or SOAR?
An alert usually represents a single signal. An incident investigation may involve multiple alerts, assets, identities, pieces of evidence, analysts, and response actions.
Case management keeps that information together. A security case may include:
related alerts;
affected assets;
users;
evidence;
investigation timeline;
severity;
analyst ownership;
response actions;
escalation records;
closure reasons.
This becomes especially important in environments operating across multiple shifts or teams. If one analyst investigates an incident overnight but the context remains in personal notes or chat messages, the next analyst may need to repeat work.
Case management provides continuity and an audit trail. Some SIEM or SOAR platforms include case management natively. Other organizations integrate with ITSM or dedicated incident platforms. The reference architecture does not require a standalone case-management product; it requires the capability.
How does an alert move from detection to response?
A typical SOC flow can be represented as:
Telemetry → Collection → Detection → Alert → Triage → Investigation → Case → Escalation → Response → Closure → Lessons Learned
Consider a suspicious authentication event.
A SIEM or XDR platform detects the activity and creates an alert. An analyst performs triage to determine whether the behavior may be legitimate. If the signal remains suspicious, the investigation expands into endpoint activity, user permissions, session history, and actions performed after authentication.
If account compromise is confirmed, the incident may be escalated and the response process activated. Sessions may be revoked, the identity disabled, and the endpoint investigated for additional evidence.
Even after containment, the SOC still needs to understand root cause, affected scope, evidence-retention requirements, and whether detection logic should be improved. The architecture therefore needs to support not only detection, but the entire path from detection to operational action.
What Does the Response Layer Include?
Response can take different forms depending on the security domain.
Domain | Example Response Actions |
Identity | Disable account, revoke session, reset credentials |
Endpoint | Isolate host, terminate process, quarantine file |
Network | Block IP/domain, modify ACL or firewall rule |
Quarantine message, block sender or domain | |
Cloud | Revoke key/token, disable access, modify policy |
Application | Disable user or restrict access |
Business | Escalation, communication, activate incident procedures |
The critical point is that not every SOC has authority to perform every action directly.
A managed SOC may only be authorized to recommend actions. An internal SOC may be able to isolate endpoints but still require approval before disabling a high-value executive account or changing production firewall rules.
Response authority therefore has to be defined as part of SOC governance.
The architecture is incomplete if it shows how alerts are generated but does not show who is allowed to act on them.
Where do reporting and governance fit?
Reporting is often placed at the top or edge of a SOC diagram, but governance influences the entire architecture.
Operational reporting may include:
alert volume;
incident volume;
backlog;
MTTD;
MTTR;
SLA performance;
false-positive rate;
recurring incidents;
detection coverage;
root causes;
risk trends.
Governance goes further. It should answer questions such as:
Who owns an incident? When must it be escalated? How long should logs be retained? Who may access security data? Which response actions require approval? How are responsibilities divided between internal teams and service providers?
This is why architecture should not only describe how tools connect. It should also describe who owns the decision when those tools identify a problem.
How should the data flow between SIEM, SOAR, and EDR/XDR be understood?
A simple reference flow might look like:
Endpoint → EDR/XDR → SIEM → Alert → SOAR → Analyst → Response
But real-world architectures are often more complex. For example: EDR → XDR Incident → Analyst while simultaneously: EDR Telemetry → SIEM
Another workflow may be: Cloud Security Alert → SOAR → Enrichment → Case Management or SIEM Alert → Analyst Investigation → XDR Response Action
The key point is that SIEM, SOAR, and XDR should not be viewed as mandatory blocks that must always appear in one fixed sequence.
They are capabilities that can interact in different ways. A more useful architecture review asks:
Where is the signal generated?
Where is the signal correlated?
Where does investigation happen?
Who makes the decision?
Which system performs the response?
Once these questions are answered, the architecture begins to reflect the actual operating model.
Does every SOC need SIEM, SOAR, and XDR?
No. A smaller SOC may rely heavily on XDR as its primary detection platform, use built-in case management, and operate without a dedicated SOAR product.
A larger enterprise with hundreds of telemetry sources and multiple cloud environments may use SIEM to centralize security data, XDR for endpoint and identity investigations, and SOAR for orchestration.
A highly mature SOC may add additional capabilities such as:
security data lakes;
threat intelligence platforms;
detection-as-code;
dedicated case management;
advanced automation;
custom analytics.
There is no universal architecture that fits every organization. The stronger design principle is:
Use Case → Telemetry → Detection → Workflow → Response
rather than: Buy Tools → Find a Use Case Later
A SOC built around the second model can easily become a collection of products without a coherent operating logic.
Does a Managed SOC use a different architecture?
Not necessarily. The core capabilities remain similar. Telemetry still needs to be collected. Detection still needs to occur. Alerts still require triage, investigation, escalation, and response.
The main differences usually involve ownership and operating boundaries. Organizations need to define:
who owns the platform;
where telemetry is sent;
where data is stored;
whether analysts are internal or external;
how much response authority the provider has;
what SLAs apply;
how critical incidents are escalated.
A Managed SOC may operate the customer's SIEM or use the provider's own platform. Some models combine SIEM, XDR, and MDR capabilities.
For a deeper comparison of managed detection models, see SOC and MDR: Navigating the differences to choose the optimal defense model
When should organizations build internally vs outsource part of the SOC?
Organizations with mature internal Security Operations teams, strong control requirements, and sufficient staffing may choose to operate most SOC layers internally.
Outsourcing may make sense when the organization needs faster monitoring coverage, lacks enough analysts, or cannot sustain 24/7 operations.
A hybrid model can also work well.
For example, a provider may handle continuous monitoring and initial triage while internal teams retain governance, response authority, and decisions involving production systems.
The key is to avoid treating outsourcing as the transfer of all security responsibility. Even with a Managed SOC, the organization still needs internal ownership, escalation paths, and defined response authority.
Organizations considering this operating model can also review SOC outsourcing
Questions to answer before designing SOC architecture
Before choosing SIEM, XDR, or SOAR products, organizations should be able to answer questions such as:
Which detection use cases matter most?
Which telemetry is required for those use cases?
Where are logs currently stored?
How long must data be retained?
Which SIEM or XDR capabilities already exist?
What is the current alert volume?
Who performs triage?
Who performs investigation?
Who has response authority?
Is a dedicated SOAR platform really necessary?
Where will case management live?
Is 24/7 monitoring required?
Are there data residency requirements?
Which SLAs need to be measured?
Who consumes operational and executive reporting?
These questions help architecture start from operating requirements rather than from a product catalog.
How can IPSIP Vietnam support SOC architecture and operations?
A SOC architecture only becomes useful when it reflects the real environment. An organization that already operates EDR, firewalls, cloud security platforms, and SIEM will have a very different design challenge from one building monitoring capabilities for the first time.
Likewise, a security team with 24/7 analysts will require a different operating model from an organization whose IT team works primarily during business hours.

A practical starting point is to review:
Current Telemetry → Priority Use Cases → Detection Platforms → Investigation Workflow → Escalation → Response Authority
For organizations that already have SIEM or XDR but are unsure whether detection, investigation, and response are connected into a coherent operating model, reviewing current data flows and workflows can identify gaps before more technology is purchased.
Businesses that need continuous monitoring, investigation, and escalation without building the entire Security Operations team internally can also explore IPSIP Vietnam’s SOC 24/7 service as a reference for how managed monitoring and escalation can be structured.
The goal is not to design an architecture with the most boxes.
The goal is to build a functioning chain: Data → Signal → Context → Decision → Action with clear technology, workflow, and ownership at every stage.
SIEM, EDR/XDR, and SOAR can all play important roles, but none of them creates a complete Security Operations capability on its own.
Organizations should begin with the use cases they need to detect, identify the telemetry required, define detection and investigation workflows, and then choose the technologies best suited to each layer.
When SOC architecture is designed around data flow, workflow, and response authority rather than product names, security teams are better positioned to turn millions of technical events into security decisions that matter to the business.
References












