What is Tachyon Agent Historic Data Capture?

On Tachyon Windows Agent devices Tachyon continuously captures events, which enables Tachyon to capture all significant events as they happen. This should be contrasted with polling, which to a certain degree relies on luck to capture conditions that are brief enough to fall between polls. In this way Tachyon Agent Historic Data Capture compares with the Windows Task Manager or Perfmon. Tachyon captures the data to a compressed and encrypted database to ensure that it has a very low impact on device performance and security.

The data is captured and stored to a local, encrypted persistent store and then periodically aggregated according to an ongoing daily, weekly, monthly window. This means that the data is held securely and the amount of data is minimized while still maintaining its usefulness.

Configuration options for each capture source are described in the public documentation reference for Tachyon Agent configuration properties.

What are the data capture sources?

The table below lists currently supported capture sources, and on which OS they are supported.

The Agent has two key mechanisms of knowing when an event occurs that is of interest - event-based and polling-based

• Event-based relies on a source external to the Agent (normally the operating system) providing a notification to indicate that something has happened
• Polling-based is where the Agent will periodically check a source of data and work out what has changed by looking at differences in the data returned

The Windows Agent can be configured to use polling instead of ETW for individual capture sources.

How do I retrieve the data from the Tachyon Agent devices?

Live and aggregated historic data is available in inventory tables. 

/* Sum the number of connections made per process today */
SELECT    SUM(ConnectionCount) AS Connections
,         ProcessName
FROM      $TCP_Daily
WHERE     TS = DATETRUNC(STRFTIME("%s", "now"), "day")
GROUP BY  ProcessName;

SELECT * FROM $Process_Live WHERE ProcessName = "chrome.exe"

Note that because the inventory tables are not created with COLLATE NOCASE, they need to be queried in a case-sensitive fashion. So the example above won't match "Chrome.exe" or "chrome.EXE" - to work around this, you can use WHERE ProcessName LIKE "chrome.exe"

How is the data managed?

The Tachyon Agent automatically aggregates and grooms data in each inventory table, according to aggregation intervals and data retention settings which are configurable in the Agent configuration file.  

• Default aggregation cycle interval is every 60 seconds, therefore it may take up to a minute before an event appears in an aggregated table
• Default retention for live tables is 5000 entries provided at least 3 aggregation cycles have occurred (older entries are deleted to make room for new entries)
• Default retention for hourly tables is a rolling 24 hours.
• Default retention for daily tables is a rolling 31 days.
• Default retention for monthly tables is a rolling 12 months.

Each aggregated table is built from the live table, and does not have a dependency on other aggregated tables. For example, Monthly is fed by Live, not fed by Daily. This allows retention settings to be configured independently for each table.

Data is stored in a local, encrypted persistent store, which persists during an Agent upgrade, uninstall and re-installation, unless specifically deleted.

If the Agent is unable to write to storage (out of disk space or other file-system problems), it will fail but continue monitoring in the hope this situation will improve later.

Historic data capture inventory schema

The following table shows the fields which exist only in the Live and Aggregated (Hourly, Daily, Monthly) tables.

Timestamps

The timestamp column (TS) in each table is stored in Unix Epoch format. Defined as the number of seconds that have elapsed since 00:00:00 Coordinated Universal Time (UTC), Thursday, 1 January 1970. 

To convert to a readable text format use the datetime SQL operation.

select Fqdn, datetime(TS, "unixepoch") as TS_ from $DNS_Hourly where Fqdn like "%facebook%";

Timestamps are truncated in the aggregated tables.

• Hourly - time is truncated to each hour - so an event that occurred at 2017-01-27 18:03:54 would be included in the summary for 2017-01-27 18:00:00
• Daily - time is truncated to midnight on each day - so an event that occurred at 2017-01-27 18:03:54 would be included in the summary for 2017-01-27 00:00:00
• Monthly - time is truncated to midnight on the first day of each month - so an event that occurred at 2017-01-27 18:03:54 would be included in the summary for 2017-01-01 00:00:00

DNS resolutions

The Agent attempts to capture DNS queries at the point that they are made, although on non-Windows platforms (and pre-Win 8.1 - see below), this is not presently possible and instead the local DNS cache is queried through polling.

When the Agent captures DNS queries, it captures the query, not the result of that query. That is, the Agent will capture a request to resolve an FQDN which may ultimately not be resolvable.

When using ETW, the Agent will not perform an initial scan to establish the contents of the DNS cache. When polling, the Agent will capture all unique FQDNs available in the DNS cache; new entries that appear in the cache are deemed to correspond to resolutions.

Process executions

On Windows, the Agent runs as LOCAL SYSTEM, therefore details of almost every process will be available; however some processes may not be accessible because of permissions.

The Agent captures process starts; it does not track how long the process has been running, or how much CPU-time (or user/kernel/active time) the process has used.

Each time the Tachyon Agent starts it does an initial scan of processes before it starts capturing. To prevent double-counting a persistent storage setting called "Inventory.ProcessesLastScan" records the last time the Agent checked for processes. This corresponds to the last time the Agent polled, or if ETW is used it is the time when the Agent inventory module was last terminated.

Software installations

Each time the Tachyon Agent starts it does an initial scan of install software before it starts capturing. To prevent double-counting persistent storage settings called "Inventory.SoftwareInstallations" and "Inventory.SoftwareInstallationsLastScan" records the results of the last scan by the Agent.  If these settings are present, the Agent will, on start-up, attempt to identify installs/uninstalls which occurred while the Agent was not capturing data. Since the Agent has no way of knowing when this install/uninstall happened, it will mark the event as having occurred "now".

On Windows, software installations are read from the registry key HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall and HKLM\SOFTWARE\Wow6432Node\Microsoft\Windows\CurrentVersion\Uninstall

Per-user installations are not yet supported.

Linux does not distinguish between O/S packages (even the kernel) and application packages; they are all software.

Outbound TCP connections

The Agent captures TCP connections, not UDP connections - as UDP is inherently connectionless (each packet sent is effectively a new connection).

Each time the Tachyon Agent starts it does an initial scan of connections before it starts capturing. A limitation of the Windows API is means that all established TCP connections, whether inbound or outbound, are captured; there is no way to distinguish between the two. This means that it is possible for the Agent to double-capture a connection if that connection was established before the Agent stops monitoring, and still exists when the Agent starts monitoring again, for example between Agent restarts. Unlike other capture sources, there is no persistent storage setting to prevent double-counting.

The Agent captures initial "connect" requests, not just successful connection establishment. This means that an attempt to perform a connection will be captured, even if that connection does not complete, for example, because of a timeout, or the server-side does not permit the connection.

datetime(TS, "unixepoch").