FortiSOAR Sizing Guide

FortiSOAR Sizing Calculator

The sizing calculator utility associated with this document helps you define your sizing requirements for FortiSOAR. This document explains how to use the sizing calculator and defines parameters such as ingestion rate, number of workflows run per day, workflow and audit purging policies, etc, required to be added in the utility. The sizing calculator utility uses specified parameter values and outputs a recommended configuration for your FortiSOAR instance.

You can download the sizing calculator from: https://help.fortinet.com/fortisoar/Sizing_LVM_Calculator_v2_0_0.xlsx.

Inputs to the sizing calculator

You need to specify the following details in the sizing calculator to calculate your FortiSOAR configuration:

1. Average number of alerts/day
2. Average number of playbooks run/day
3. Playbook logs retention policy in weeks (recommended 52 weeks)
4. Audit logs retention policy in weeks (recommended 52 weeks)
5. Whether Threat Intelligence Feeds are enabled (default is Not Enabled).

Other defaults in the sizing calculator

The following additional default details in the sizing calculator need to be specified to calculate your FortiSOAR configuration:

1. Disk size computation:
   1. Primary Data: For every alert, the calculator considers 1 MB of primary data to be generated. This is an approximate number considering:
      1. 8 indicators extracted
      2. 10 comments added, including one attached file of approximately 0.5 MB.
         Note: If your investigation relies on heavier attachments or screenshots, or primarily relies on email ingestion with large images, you might consider doubling the disk size projections. Refer to the “Test Run” section, which considers an - additional large attachment approximately ~0.5 MB in size that is uploaded as a comment, such that ~1 MB of primary data gets generated in the environment for every alert ingested
   2. Audit Logs: The calculator considers around 7 GB of audit data to be generated weekly.
   3. Workflow Logs: The calculator considers log size of 5KB generated per playbook.
      You can run the following command on your FortiSOAR instance to confirm the database consumptions for your current data and change the inputs to your sizing calculator accordingly:
      csadm db –getsize
      

      The above command gives the total database sizes. Size per playbook log could be obtained by dividing the total ‘Workflow Logs’ size from the above output by the total ‘Executed Playbook Logs’ in the UI. To get the size per alert/incident, divide the total ‘Primary Data’ size by total number of alerts in the UI.

   4. Feed Data: The calculator considers feed data for approximately 500,000 feeds as approximately 10 GB.
2. CPU and Memory based on playbooks run/day used by the Calculator:
   

   Playbooks run/day	Configuration
   Up to 10,000	16GB RAM, 4 core CPU
   10,000 – 50,000	32GB RAM, 8 core CPU
   50,000 – 150,000	64GB RAM, 16 core CPU
   150,000 – 250,000	Active/Active HA cluster with 32GB RAM, 8 core CPU

   These sizes are recommended keeping in mind long term sustenance and average workflow execution times of 15 seconds/workflow. Your environment might also have higher or lower scale limits based on the workflows runs.
   

   Playbook runs involve frequent disk I/O. Having SSD disks with a higher guaranteed IOPS (2000 or higher) are strongly recommended in the production environment for the best performance.

LVM Sizing Calculator

The LVM sizing calculator utility associated with this document helps you define your sizing requirements for FortiSOAR when you disk size is not as per the standard configuration of 300GB. You can use the LVM sizing calculator to calculate the sizes of partitions to be created according to your disk size.

You can download the sizing calculator from: https://help.fortinet.com/fortisoar/Sizing_LVM_Calculator_v2_0_0.xlsx.

Inputs to the LVM sizing calculator

To calculate the sizes of partitions to be created according to their disk size, users only require to enter the disk size in cell D22 in the 'LVM Sizing Calculator' section of the Sizing Calculator.

Test Run

The following sections help you to further understand the sizing calculation logic with the help of results from a sustenance run. The tests were run on the default recommended hardware configurations and using a common daily ingestion volume seen in customer environments. It shows details of the system utilization over the period of the run. The test results can be used as a reference for deciding on the CPU, memory, and disk of your FortiSOAR instance.

Test Configuration 1

For each of these tests the load varies in terms of the number of alerts ingested per day. The following parameters are common for each of the runs.

Instance Configuration:

FortiSOAR Virtual Appliance Specifications

Component	Specifications
CPU	16 CPUs
Memory	30 GB
Storage	500 GB virtual disk, HDD type gp3 with IOPS 3000,
Instance Type	VMWare

Operating System Specifications

Operating System	Kernel Version
Rocky Linux 8.8	4.18.0-477.13.1.el8_8.x86_64

Load:

• ~5040 alerts/day (2 schedules are run: one creates 1 alert every minute, and the second creates bursts of 150 alerts every hour)

Default use-cases run per alert (9):

• SLA Calculation (All applicable SLA Playbooks)
• Alert Assignment Notification
• Indicator Extraction
• Enrichment
• Triage
• User Assignment
• Computing Alert Priority
• Alert resolved date and assigned date updated
• Alert Investigation Playbooks

Record sizes:

• 8 Indicators are created per alert.
• Each alert has 10 small text comments, 2 comments with a screenshot and a large 500KB File Attachment as a comment. Around 1 MB of primary data gets generated per alert ingested.
• Sample alert data: You can download this sample alert data from: https://help.fortinet.com/fortisoar/Sizing_Alert.zip.

Audit log and Work log retention

• Audit log retention: 7 days
• Workflow log retention: 7 days

Other FortiSOAR Tunables

The following configurations were updated as recommended for the production instance:

• Workflow workers: 16
  /etc/celery/celeryd.conf: CELERYD_OPTS="--concurrency=16"
• Postgres shared buffer: 2GB
  /var/lib/pgsql/12/data/postgresql.conf: shared_buffers = 2048MB
• ElasticSearch Xms and Xmx 8GB:
  /etc/elasticsearch/jvm.options.d/fsr.options:
  -Xms8g
-Xmx8g

Results

Data Disk Consumption:

1. Postgres partition consumption: /var/lib/pgsql
2. Elasticsearch disk consumption: /var/lib/elasticsearch
   

Time Span	Primary Data Size	Audit Logs Size	Workflow Logs Size	Elasticsearch Size	Total Disk Size
After 1 week	31GB	11.2GB	14.4GB	2GB	58.6GB
After 2 weeks	62GB	24.2GB	28.8GB	4GB	119.2GB
After 4 weeks	124GB	44.8GB	57.6GB	8GB	234.4GB
After 1 year (projected)	1612GB	582.4GB	748.8GB	104GB	3047.2GB

Note: Total Disk Consumption is calculated as Primary data size + Audit log size + Workflow logs size + Elasticsearch size + Feed data size.

Test Configuration 2

The configuration for this test is the same as Test Configuration 1, apart from the record sizes of the alert ingested.

Record sizes:

• 8 Indicators are created per alert.
• Each alert has 10 small text comments, 2 comments with a screenshot. Around 0.5 MB of primary data gets generated per alert ingested.
• Sample alert data: You can download this sample alert data from: https://help.fortinet.com/fortisoar/Sizing_Alet_Without_Attachment.zip.

Results

Data Disk Consumption:

Time Span	Primary Data Size	Audit Logs Size	Workflow Logs Size	Elasticsearch Size	Total Disk Size
After 1 week	8.2GB	11.2GB	15.5GB	1.9GB	36.8GB
After 2 weeks	16.4GB	22.4GB	31GB	3.8GB	73.6GB
After 4 weeks	32.8GB	44.8GB	62GB	7.6GB	147.2GB
After 1 year (projected)	426GB	582.4GB	806GB	98.8GB	1913.2GB

Note: Total Disk Consumption is calculated as Primary data size + Audit log size + Workflow logs size + Elasticsearch size + Feed data size.

Test Configuration 3

Apart from the instance configuration and load, all the other configuration for this test is the same as Test Configuration 1.

Instance Configuration:

• Same configuration as mentioned in Test Configuration 1 + a cluster of two FortiSOAR machines that are joined in the Active-Active state using the FortiSOAR HA feature.

• The machines that form the HA cluster must be in the same network subnet.

Load:

• ~9360 alerts/day (2 schedules are run: one creates 4 alerts every minute, and the second creates bursts of 150 alerts every hour)

Results

Data Disk Consumption:

Time Span	Primary Data Size	Audit Logs Size	Workflow Logs Size	Elasticsearch Size	Total Disk Size
After 1 week	57GB	20.5GB	43.6GB	3.5GB	124.6GB
After 2 weeks	114GB	41GB	87.2GB	7GB	249.2GB
After 4 weeks	228GB	82GB	174.4GB	14GB	498.4GB
After 1 year (projected)	2964GB	1066GB	2267GB	182GB	6479GB

Note: Total Disk Consumption is calculated as Primary data size + Audit log size + Workflow logs size + Elasticsearch size + Feed data size.