Eon Threat Center brings threat detection and recovery to the snapshots you already take. It reads every protected copy in scope for ransomware, malware, anomalous change, AI data risk, and vulnerable applications. Then it holds the last clean copy of each resource before your retention policy deletes it.
It does this without touching production. Scanning runs on Eon’s own compute against data already at rest. There’s nothing to install on your resources.
During an incident, teams need to identify the threat, find safe recovery points, and restore hundreds of resources under pressure. Eon Threat Center brings these steps together, helping teams move from investigation to restoration with confidence.
With Eon Threat Center, teams can:
- Detect file ransomware by behavior, not signature: Inspect file structure and measure entropy against each resource's previous snapshot and flag mass renames, ransom notes, mass deletion, malware and revoked encryption keys. Covers VMs, disks, object and file storage, plus drive content in Microsoft 365 and Google Workspace.
- Catch database attacks that file scanning misses: Read PostgreSQL, MySQL, SQL Server, Oracle, DynamoDB, MongoDB, and Keyspaces logically to find dropped tables, collapsing row counts, columns encrypted in place and ransom demands written into rows.
- Get one verdict, not a queue of alerts: Each finding carries a conclusion, a confidence figure, a plain-English reason, the exact file path or table, and a risk rating weighted by what the resource holds.
- Hunt with your own detection rules: Run Yara rules across your resources backups or search for objects globally using SHA-256 hash.
- Recover hundreds of resources from verified clean copies: Restore up to 500 resources per batch, each from its own clean snapshot and restore template, re-checked before execution and gated by multi-party approval.
A green job is not a clean copy
The protection job went green. That tells you a copy was made, but nothing about what is inside or whether the backed up files are encrypted or corrupted
Attackers count on that gap. They encrypt in place, delete quietly and wait. Every snapshot after the compromise carries the damage forward, and retention keeps aging out the healthy copies behind it. By the time anyone looks, the last clean recovery point may already be gone.
Your EDR and SOC tools watch the live estate. None of them can see the history held in your protected data. So the question that matters most in an incident, which copy is safe to restore, usually has no fast answer. That's the gap Eon Threat Center closes.
How Eon Threat Center works
Scope detection the way you scope protection
Detection is assigned exactly the way protection already is. Inherit conditions from your existing protection policies, or target resources by data classification, environment or tags. A resource created next month is scanned because it matches, not because someone remembered to add it.
Read files for what ransomware leaves behind
Encrypting a filesystem is loud. Readable content turns to noise, file structures become malformed, files are replaced en masse by a different type, and ransom notes appear.
Eon checks these signals against each resource’s history, rather than a list of known strains. It measures entropy in each file’s opening bytes and checks its structure to confirm that a PDF is still a PDF and a spreadsheet is still a spreadsheet. A known file type becoming unrecognizable between snapshots is one of the strongest signals.
To distinguish attacks from normal churn, Eon runs a chi-square test on every folder, comparing the mix of file types that appeared with those that disappeared. Documents and spreadsheets replaced one for one by a single unknown type register as statistically unlikely to be ordinary work.
Before reaching a verdict, Eon considers the resource, the data it holds, and the directory. Files in a user’s documents carry different context from those in cache or system paths that churn by design.
The same engine covers OneDrive, SharePoint, and Google Drive. Scanning runs on Eon’s compute against the secondary copy, surfacing tenant-wide encryption events with real file paths and no calls against your tenant.
Read databases the way databases fail
A database under attack doesn’t look encrypted. Its files are already dense binary, normal operations rewrite them constantly, and a ransom note in a row is invisible on disk. File scanning walks past all of it.
Eon learns each database’s schema and history to forecast what the next backup should look like, with upper and lower boundaries. With fewer than ten snapshots, it stays in learning mode without alerting. Missing backups are extrapolated from the pattern, and seasonality is learned, not assumed. In other words, when weekend backups dip, the boundaries dip with them.
Comparing each backup against that forecast reveals vanished tables, row counts below expectations, changes outside the normal rhythm, and sudden spikes in distinct column values.
For every column, Eon tracks distinct values as a ratio of rows, so the signal behaves consistently across ten thousand rows or ten billion. A status column that jumps from five values to a unique value in every row can reveal encryption in place, even when the row count hasn’t changed. Expected churn in columns such as timestamps and versions doesn’t count against you.
Not every big change is ransomware. The difference shows up in the data:
- Ransomware encrypts what it doesn’t delete. Values become random: a column containing a few hundred country codes suddenly holds a distinct value in every row. Eon calls this a cardinality explosion. Attackers also commonly drop tables and leave ransom notes in new tables or existing rows.
- A bad migration can rename, split, or drop tables, but the values remain recognizable. Country codes are still country codes.
- Human error or an insider wiping data deletes tables or rows while leaving surviving values untouched and no ransom note. Eon flags this as a data anomaly rather than scoring it as ransomware.
Key changes follow the same logic. A new KMS key within your account or organization is routine. A revoked key, or one held somewhere you no longer control, signals a potential attack.
Turn evidence into one verdict
Evidence is weighted, not counted. A ransom note carries more than an entropy shift alone, and two weak pieces of evidence together can outrank one strong one. What comes out is one conclusion: clean, ransomware, malware, anomalous activity, AI data risk or data exposure, with a confidence figure, a written reason and a risk rating.
Both engines use the same conclusion model, so file and database findings land on one board, scored the same way. Filter by detector, risk, path, table or database. A day-by-day view across your retention window shows where findings begin, which marks the edge of your compromise window.
Findings route by type and condition to your SIEM, Slack or webhook. When a mass change is yours, like a key rotation or a planned migration, exclude it by evidence and detector, scoped to one resource or the whole organization.
Hunt with your own rules
Eon Threat Center runs YARA against the contents of your protected data: curated rule packs plus rules you write in the console, validated as you write them. Queue a hunt and it runs with the resource's next snapshot. Every completed hunt records the rules in force, the files read and skipped, and what matched. You are also able to search globally across all resources using SHA-256 to see where else the threat lives.
From verdict to recovery
Detection is only worth something if it ends in a restore you can trust. Eon Threat Center carries every verdict through to that point.
Hold the last clean copy
When a threat lands on a resource, the deletion hold moves to that resource's last fully clean snapshot. Retention keeps aging out of everything else. The copy you need stops being something you have to notice and rescue by hand.
When no clean snapshot remains, Eon holds nothing rather than the most recent one. Holding an infected copy would present it as safe, and turn an honest "we have no clean copy" into a restore that reinfects production.
Recover hundreds of resources as one operation
An incident is never one resource. The Remediation Center opens with the number of resources currently flagged for ransomware, resolved from live findings rather than a hand-maintained list, and one button. From there, each resource gets:
- Its own clean snapshot, resolved inside a lookback window you choose. Resources with no clean copy in the window are reported separately.
- Its own restore template, so it comes back configured the way it was built. Resources with no compatible template are dropped.
- A last check before execution, confirming the snapshot is still clean. Anything carrying findings is dropped, never restored.
- An approval gate, where multi-party approval is in force. Submitting raises a request; the screen says pending, not started.
Each batch covers up to 500 resources. Every step can drop a resource, because a bulk restore that proceeds on something it couldn't verify is how you reinfect production at scale.
Restore templates belong to your disaster recovery drill, when nothing is on fire. Eon revalidates every template against your current infrastructure and flags the ones a retired subnet or deleted security group has broken, so the work from the spring drill still holds in the autumn incident.
Remediated means clean, not finished
A restore creates new resources. The infected originals still hold the payload and still need isolating through your incident-response process, and Eon says so rather than letting a green job imply a clean environment. A resource counts as remediated only once it has been protected and rescanned clean.
Built for the agents you already run
Eon’s remediation agent investigates incidents and builds recovery plans, identifying affected resources, what was hit, when the compromise window opened, and which resources contain PII or PHI and should be recovered first. For each resource, Snapshot Architect can automatically assemble the best recovery recipe using the same clean-snapshot resolution, templates, and approval gate as the Remediation Center. The agent then waits for your approval.
Threat Center is built for agents, with every operation available over MCP. Your own agent, whether Claude or Cursor, acts on your behalf with your permissions and the same approval gate.
Get started with Eon Threat Center
Eon Threat Center is available today alongside Eon Data Protection. To learn more, visit the Eon Threat Center or request a demo to see it in action.




