Skip to main content

Data Aging

When working with data at rest, two big topics that arise from a security operations perspective:

  1. Manage retention span to maximize retrospective detection capabilities
  2. Comply with data residency requirements and privacy regulations

Often, these two topics conflict each other: security teams want maximum retention spans to find advanced attacks retroactively once intelligence becomes public. The recent watershed events, like SolarWinds and log4j, demonstrate that post-hoc detection must now span months to years to successfully uncover deep infiltrations. Conversely, security data contains often personal information and falls under strict regulations that prohibit unfettered use. There exists a strict upper bound on storing specific security event data for retrospective detection.

Data Constraints Data Constraints

How can we implement a successful detection and response strategy in this spectrum? By making the spectrum transparent and fully controllable. In VAST, we developed a declarative compaction approach to perform fine-grained transformation of historical data to control retention span and manage finite storage. Compaction operates in two dimensions:

  1. Temporal: temporal compaction defines what to do with data as a function of age. For example, a policy may dictate a maximum retention of 1 week for events containing URIs and 3 months for events containing IP addresses related to network connections. However, these retention windows could be broadened when pseudonomyzing or anonymizing the relevant fields. VAST not only supports deletion of data after exceeding a configured age, but also transforming the data with a set of transformations (such as hashing, encrypting, permuting). The intuitive declarative definition makes it easy to express data residency regulations in a shareable form, e.g., with data protection officers.

  2. Spatial: Traditionally, reaching a storage budget triggers deletion of the oldest (or least-recently-used) data. This is a binary decision to throw away a subset of events. What if you could only throw away the irrelevant parts and keep the information that might still be useful for longitudinal investigations? What if you could aggregate multiple events into a single one that captures valuable information? Imagine, for example, to halve the space utilization of events with network flow information and keeping them 6 months longer; or imagine you could roll up a set of flows into a traffic matrix that only captures who communicated with whom in a given timeframe.

    By incrementally elevating data into more space-efficient representations, spatial compaction gives you a much more powerful mechanism to achieve long retention periods while working with high-volume telemetry.

Key Benefits

In summary, VAST's data aging capabilities have the following benefits:

  • Easy Compliance: VAST makes it easy to implement and share compliance policies. The configuration of temporal compaction is a human-readable file that clearly states what happens with the data at what age.

  • Gradual Event Decay: instead of deleting old data as a whole, VAST offers an incremental approach to reduce size and information content of events. For example, first filter out unneeded fields, then aggregate the smaller events into an event summary, and delete the summary at last. Being able to express event decay in this gradual way is the Goldilocks approach of managing utility of security telemetry.

  • Dynamic Aging: When only considering age as input for deleting old data, high-volume data source also dictate the retention span for low-volume event streams, because they can consume orders of magnitude more space. VAST's weighted aging makes it possible to define a relative importance of events to each other. By assigning higher age weights to low-volume-but-important data sources (e.g., alerts), we can selectively increase their retention span. This unique age prioritization makes it easy to define importance of events independent of the event mix.