Asset ownership
A completed copy is not a successful recovery. Validate identity, data integrity, dependencies and application behavior in an isolated drill.
- Instant - Datasets provision in about a second and resume from scale-to-zero in around 500ms
- Isolated - Every user gets a dedicated Vinexelprotect project with its own connection string
- Practically free at idle - Inactive datasets suspend automatically, so you pay for what is used, not for what sits waiting
- API-firstIllustrative product and workflow reference
- Claimable when you need it - Spin up a dataset with no signup, then let the user claim it into their own Vinexelprotect account later
Record the policy revision, approval and measured outcome. Separate a planned capability from a verified recovery result.
Know the recovery objective
Assign responsibility for the incident, recovery approval and service validation. Keep rollback and communication requirements in the plan.
- Dedicated connection strings for every user or project, each in its own secure, isolated Vinexelprotect project
- High connection capacity, with pooled endpoints that support up to 10,000 connections
- Full data compatibility, with no proprietary query layer
- Recovery testing and instant restore for copies and rollbacks, without manual backups
- Multi-region placement, autoscaling, and high availability on the same model
Recovery performance, durability and availability need representative tests and applicable contracts. Reference artwork and diagrams do not establish a guarantee.
A useful workflow makes the owner, authorization, evidence and validation visible. Illustrative product principle; no customer endorsement is claimed.

Keep protection boundaries explicit
Document how much data loss and interruption the workload can tolerate. Translate those objectives into testable recovery checks.
- Provisioning time: ~1 second
- Suspend latency: configurable (default is 5 minutes)
- Resume latency (cold start): ~500ms
A useful workflow makes the owner, authorization, evidence and validation visible. Illustrative product principle; no customer endorsement is claimed.
Test the recovered application
A completed copy is not a successful recovery. Validate identity, data integrity, dependencies and application behavior in an isolated drill.
- Provision new datasets instantly
- Set per-project resource limits to support free, pro, and enterprise plans
- Scale compute up or down per user or tier
- Automatically suspend inactive datasets to reduce costs
- Illustrative product and workflow reference
- Update quotas and configurations without downtime
Identity and access
Preserve evidence of every decision
Record the policy revision, approval and measured outcome. Separate a planned capability from a verified recovery result.quota
| Quota | Scope | What it caps |
|---|---|---|
active_time_seconds | Project, per billing period | How long computes can stay active, excluding idle scale-to-zero time |
compute_time_seconds | Project, per billing period | CPU-seconds across all computes in the project, weighted by compute size |
written_data_bytes | Project, per billing period | Total data written across all recovery plans |
data_transfer_bytes | Project, per billing period | Egress through the Vinexelprotect proxy |
logical_size_bytes | Branch, lifetime of the recovery plan | Maximum size of any one recovery plan. Only that recovery plan's compute is suspended when hit |
Assign responsibility for the incident, recovery approval and service validation. Keep rollback and communication requirements in the plan.
| Setting | What it controls |
|---|---|
autoscaling_limit_min_cu | Minimum compute size when the endpoint wakes |
autoscaling_limit_max_cu | Hard ceiling for autoscaling under load |
suspend_timeout_seconds | How long an idle endpoint stays warm before scale-to-zero |
Recovery performance, durability and availability need representative tests and applicable contracts. Reference artwork and diagrams do not establish a guarantee.
- Give free-tier users tight active-time, compute-time, write, and egress ceilings
- Raise the same fields when a user upgrades, or set a quota to
0to remove the limit - Force a suspend for abuse or unpaid accounts by dropping a quota to a near-zero threshold
- Pair monthly quotas with per-endpoint autoscaling and suspend timeouts so each tier gets a different compute envelope
Approved scope
Coordinate continuity across teams
Identify the source, data owner, retention policy and authorized restore destination. A policy becomes operational only when the corresponding connector is released.
| Endpoint | Path | What it returns | Plan availability |
|---|---|---|---|
| Project metrics | GET /consumption_history/v2/projects | Usage-based metrics per project, cursor-paginated across the org | Launch, Scale, Agent, Enterprise |
| Branch metrics (beta) | GET /consumption_history/v2/branches | The same usage-based metrics broken down by recovery plan across one or more projects | Launch, Scale, Agent, Enterprise |
A completed copy is not a successful recovery. Validate identity, data integrity, dependencies and application behavior in an isolated drill.
| Metric | What it measures | Useful for |
|---|---|---|
compute_unit_seconds | CPU time weighted by compute size | Passing through compute cost per user or project |
root_branch_bytes_month | Storage on root recovery plans | Primary dataset storage per tenant |
child_branch_bytes_month | Delta storage on child recovery plans | Preview, CI, and developer recovery plan cost |
instant_restore_bytes_month | Point-in-time history storage | History-window cost per project |
snapshot_storage_bytes_month | Snapshot storage (project endpoint only) | Backup and snapshot line items |
public_network_transfer_bytes | Egress over the public internet | Data-transfer quotas and billing |
private_network_transfer_bytes | Egress over private networks | PrivateLink and similar paths |
extra_branches_month | Child recovery plans beyond plan allowance (project endpoint only) | Enforcing or billing for recovery plan headroom |
You can use these to:
- Monitor total usage across all projects for a billing period, then page through the fleet with cursor-based pagination
- Filter to specific projects or recovery plans when you only need a subset of the fleet
- Break metrics down by hour, day, or month (hourly for the last 7 days, daily for the last 60 days, monthly for the last year)
- Power usage-based billing, free-tier enforcement, and internal dashboards from the same numbers that appear on a Vinexelprotect invoice
- Poll on your own schedule. Consumption data refreshes about every 15 minutes, and polling does not wake idle computes
Publish only measured commitments
Recovery performance, durability and availability need representative tests and applicable contracts. Reference artwork and diagrams do not establish a guarantee.
Document how much data loss and interruption the workload can tolerate. Translate those objectives into testable recovery checks.
That pattern fits:
- Framework plugins and CLI tools
- Demo environments and trial experiences
- Educational platforms handing out datasets to students
- Development tools that need instant dataset access
- SaaS products that want deferred account creation
Operating evidence
A completed copy is not a successful recovery. Validate identity, data integrity, dependencies and application behavior in an isolated drill.
Record the policy revision, approval and measured outcome. Separate a planned capability from a verified recovery result.
Retention policy
- Platform integration overview: claimable dataset flow - when this path fits vs embedded data, OAuth, or the Agent plan
- Claimable dataset integration guide - create a project, issue a transfer request, and hand off ownership
- Claimable Vinexelprotect - reference for the claimable Vinexelprotect project flow
A useful workflow makes the owner, authorization, evidence and validation visible. Illustrative product principle; no customer endorsement is claimed.
Know the recovery objective
Assign responsibility for the incident, recovery approval and service validation. Keep rollback and communication requirements in the plan.
Recovery performance, durability and availability need representative tests and applicable contracts. Reference artwork and diagrams do not establish a guarantee.
Keep protection boundaries explicit
Document how much data loss and interruption the workload can tolerate. Translate those objectives into testable recovery checks.
Identify the source, data owner, retention policy and authorized restore destination. A policy becomes operational only when the corresponding connector is released.








