Agentless log collection refers to gathering log data from various sources without installing dedicated software agents on the systems generating the logs. Instead, it leverages protocols such as SNMP traps, WECS, WMI, and syslog to retrieve log data remotely.
It is easier to explain what agentless log collection is by also providing some context about agent-based log collection. The truth is that these two options for collecting logs walk hand in hand, meaning that they can and will likely coexist on your network. Interestingly, when people talk about this topic, it often turns into a debate about which log collection method is better. Here comes the spoiler: neither method is better than the other, and to give an extra twist to the story, this isn’t even an entirely valid statement. It is really like comparing apples and oranges.
I would like to emphasize that whenever possible, you should always use an agent for collecting logs, as an agent-based log collection method almost always brings more to the table. Agent-based log collection provides more features, allows for filtering, trimming, and parsing right at the source, and generally has much more benefits. But, there is always a but…
Why and when to use agentless log collection?
There are scenarios where installing a log collection agent is not possible, for various reasons.
In the case of network devices, such as hubs, routers, firewalls, switches, or networked printers, the operating system is embedded. So, installing any third-party software is simply impossible, as the operating system cannot accommodate that. There are also scenarios where requirements may be in place that prevent the installation of a log collection agent, such as security or mandatory compliance reasons. In these cases, there is no other choice but to opt for agentless log collection.
However, this does not mean you must eliminate a versatile log collection agent, as it still has an important role further down the chain.
Do you still need an agent?
Well, yes! Especially if you use NXLog, as it provides an agent that does more than just collecting logs. You can use it as a log aggregator and log relay, which as I mentioned, becomes very important in the journey of your log records. Just because you do not use an agent on your end devices (where the logs are coming from), it does not mean that they will send the logs directly to your SIEM. This is something that you should try to avoid at all costs:
-
Configuring many devices to directly send log data to your SIEM quickly becomes cumbersome to manage.
-
It can seriously jeopardize your IT security if you have a cloud-based SIEM, since that requires opening up your network to the cloud from many devices.
-
You should always have your logs enter their final destination already filtered, trimmed, parsed, and noise-free. With NXLog, if you use agent-based log collection, this is done at the source. If you use agentless log collection, where at best only basic filtering can be done at the source, you should perform this as a next step.
In the case of agentless log collection, the journey of the log data should not differ. The only difference is how the logs get to the "first hop", using a networking analogy.
The diagrams below outline how this works in a simplified way.
In the diagram, NXLog agents are installed on all nodes and are responsible for filtering, parsing, trimming, and clearing out all unnecessary noise. Then, only the necessary and valuable log data enters the network, heading to a central NXLog relay server. In a typical scenario, this central server aggregates all your logs and directs them to their final destination, or often times multiple destinations. If the destination of your logs is in the cloud or outside your local network, then this is the only machine with a connection to the internet, which has significant security benefits.
The flow in the second diagram is the same, except that no agents are installed on the nodes. The nodes send log records using various protocols to the central NXLog server. In this case, the log records are mostly unfiltered, without trimming, parsing, or any normalization.
This is the point where the features of NXLog become very handy, as you need to get your log data in the best possible shape, the very latest at this point.
Importance of log aggregation
Well, adding an extra step to the process sounds like complicating the journey of your logs, but you should always perform log aggregation. In addition to the previously mentioned security benefits, aggregating your logs to a central relay poses many other benefits:
-
You can separate the route your different logs take, for example:
-
Forwarding your database logs to destination A and your Domain Controller logs to Destination B.
-
Forwarding all your logs with a certain severity to destination C and the remaining ones to destination D.
-
Forwarding any of your logs to multiple destinations at the same time.
-
…or literally whatever requirement you might have, since NXLog supports as many routes as needed.
-
-
This is the only machine that needs an internet connection.
-
All your logs are streamlined to your SIEM or analytics solution, considerably simplifying the integration between the systems.
-
In the case of agentless log collection, the aggregation step is the best place to filter, trim, parse, and generally normalize your logs before reaching your SIEM or analytics solution. In itself, this has many benefits, from reducing network traffic to having better analytics with easier visibility of malicious activity.
Takeaways
After understanding what agentless log collection is, let’s recap the following takeaways:
-
Agentless log collection and agent-based log collection are not mutually exclusive, just suited to different use cases. In the wild, you’ll need to leverage a mixture of agent-based and agentless log collection techniques to onboard disparate log events into your SIEM or forward them to other types of destinations.
-
Agentless log collection is best suited for network devices and any scenario where an agent cannot be installed.
-
In the case of agentless log collection, the logs follow the same route as they would follow in agent-based log collection.
-
In the case of agentless log collection, filtering, trimming, and normalization are done a stage later, as they cannot be done at the source.
-
Log aggregation is still an integral part of the log data journey when using an agentless log collection method.
Finally, be sure to also check out our in-depth comparison of the agent-based and agentless log collection strategies.
