Getting Started with HCL BigFix: A Comprehensive Guide
By: Casey Cannady : IBM Certified BigFix practitioner & enterprise architect
TL;DR
BigFix is a server, a database, a layer of relays, and one small agent on every endpoint. The agent does the thinking locally and only reports what matters, which is why it scales. Before you install anything, get three things right: protect the private key that gets generated with your license (anyone holding it and its password controls every managed machine), open port 52311 for TCP and UDP everywhere inside your network, and plan relays before clients. Then start with read-only visibility, prove every action on a pilot group, and let automation come later.
I have been deploying BigFix since before HCL owned it, first as the in-house subject matter expert at Kroger Technology and later as a consultant and enterprise architect for Fortune 500 clients. The platform is one of the most capable endpoint management tools on the market. It is also easy to stand up badly, and the mistakes you make in the first week tend to follow you for years.
This is the guide I wish every new BigFix admin got on day one. It is not a replacement for HCL's installation documentation. It is the map you should have in your head before you open it.
How BigFix Actually Works
Most management tools poll endpoints from the center. BigFix flips that. Each agent evaluates content against its own machine and only reports back what is true. The server publishes content, the agent decides whether it applies, and the network only carries the answers.
That content comes as Fixlets and Tasks. Each one carries Relevance statements that describe when it applies to a computer, for example a specific operating system with a specific version of a program installed. When an operator wants to apply one, they create an action and target it at the computers where it is relevant, or at a specific list.
Relevance is the query language underneath all of it. You do not need to write it on day one. You do need to know it exists, because every serious BigFix admin eventually lives in it.
The Components
| Component | What it does |
|---|---|
| BigFix Server | Gathers Fixlet content from the internet and distributes it to relays. Every deployment has at least one. |
| Database | Stores everything the server knows. Microsoft SQL Server on Windows; on Red Hat Enterprise Linux, IBM Db2 or Microsoft SQL Server. Oracle is not an option. |
| Relays | Share the server's load and cache content close to the endpoints. HCL's rule of thumb is one relay for every 500 to 1,000 computers. |
| Clients (agents) | Run on each managed endpoint, evaluate Relevance locally, report results, and execute actions. |
| Console | The operator application. It connects to the server and keeps its view of your network current. |
| Web Reports and WebUI | Browser-based reporting and a browser-based operator experience, installed alongside the server. |
Sources for the table: HCL's basic deployment, server requirements, and database requirements pages. Check the current support matrix before you pick versions; it changes with every release.
Before You Install: The Key That Controls Everything
Installation starts with a license authorization file. From it, the installer generates your license files and the masthead: a public key (license.crt), a private key (license.pvk), and a masthead file that carries your deployment's configuration, license, and security information, including where trusted content comes from.
HCL's documentation is blunt about it: anyone with the private key file and its password has full control over every computer with a BigFix client installed. Treat it like the keys to the whole company, because it is. Store it offline, back it up in more than one secured location, and limit who knows the password. A lost key is not a support ticket. It is a rebuild.
The Network Rule That Decides Whether It Works
By default, BigFix uses port 52311 for every component, console included. HCL's network configuration requirements say TCP and UDP on that port must be completely unblocked at internal routers and firewalls. You can disable UDP, but relays use small UDP packets to tell clients there is something new to fetch, and without them clients find out later.
In my experience, most “BigFix is slow” complaints in a new deployment are really a firewall rule somebody forgot, not a platform problem. Get the network team in the room before install day, not after.
Plan Relays Before Clients
- Map your sites and link speeds first. Relays belong where endpoints cluster and where WAN links are thin.
- Size with HCL's rule of thumb, one relay per 500 to 1,000 computers, then adjust for bandwidth and how much content you push.
- Keep the hierarchy flat and boring. A few top-level relays under the server, child relays under those. Clever hierarchies are hard to troubleshoot at 2 AM.
- Plan for endpoints that leave the building. Laptops on home networks need a path back, usually a relay placed in a DMZ.
Getting Agents onto Endpoints
HCL ships a Client Deploy Tool that pulls a computer list from Active Directory or a file you provide, checks whether the client is already installed, and pushes it to Windows, UNIX, and Mac targets. Plenty of shops also bake the agent into their OS images or push it with the software distribution tool they already run. Pick one method per platform and document it; mixed install methods become mixed client versions faster than you would think.
Your First Thirty Days
These are my rules, not HCL's. They come from cleaning up a lot of deployments that skipped them.
- Look before you touch. Activate analyses and build reports first. Knowing what is actually on your endpoints is the win that pays for everything else, and it cannot break anything.
- Build a pilot group, then use it every time. Computer groups can be automatic, with membership decided by Relevance. A pilot group that represents your real hardware and software mix is the cheapest insurance you will ever buy.
- Do not run everything as a master operator. Give people the roles and computer scope they need. A console that anyone can target at everything is an incident waiting for a bad day.
- Name things like you will forget why you built them. Groups, baselines, custom sites, and actions all need names a stranger can read. You will be that stranger in six months.
- Close what you open. Stop and delete actions that are done. Hundreds of stale open actions slow the console and bury the ones that matter.
When You Are Ready for More
Once the basics run clean, the real value starts: policy actions that heal drift on their own, content managed like code, and the REST API feeding the rest of your stack. I wrote about that in BigFix Automation: Beyond the Basics. And if you are wondering why BigFix still means this much to me, that story is in Why is my favicon still the classic BigFix logo?
Standing up BigFix and want a second set of eyes?
Architecture reviews, relay planning, and cleanup of deployments that grew without a plan.
Sources & Further Reading
- HCL Product Documentation: Basic deployment, the source for the component roles and the 500 to 1,000 computers per relay rule of thumb.
- HCL Product Documentation: Server requirements and Database requirements, the source for supported operating systems and databases.
- HCL Product Documentation: Managing licenses, the source for the license files, the masthead, and the private key warning.
- HCL Product Documentation: Network configuration requirements, the source for port 52311 and the TCP and UDP guidance.
- HCL Product Documentation: Introducing Fixlets and Tasks, Introducing Computer Groups, and Using the Client Deploy Tool.
- BigFix Developer: Relevance guide, the place to start learning the query language.
Sourcing note: platform facts come from HCL's BigFix 11 documentation and are linked where they appear. The advice in “Your First Thirty Days” and the relay planning list is my own professional opinion from years of deployments, not HCL guidance. Supported versions change with every release, so check the current support matrix before you install. Revised September 2026: an earlier version of this guide incorrectly listed Oracle as a supported database and described web console ports that do not match the documentation; both are fixed.
Connect with Casey
Have a story, a question, or want Casey to write about a specific topic? DM me and tell me which story you want next.
| Websites | |
| Threads | |
| Bluesky | |
| YouTube |
Casey writes about endpoint management, cybersecurity, nomadic life, and navigating the world as a late-diagnosed AuDHD adult. New posts drop on my professional website.