Concept

EBA Smart Transport

The SNA-side smart transport role that sends protected data to SHA and uses extent maps from snagent when available.

ReadyCurrenteba-sttransportcbtsna

Definition

EBA-ST is EBA Smart Transport. It is the SNA-side Sendense data path for reading protected workload data and sending it to SHA over the secure tunnel.

EBA-ST runs as part of the SNA data path. It is a capability provided by the SNA, not a separate appliance to deploy or manage.

Backup Data Path

EBA-ST reads protected workload data on the SNA using the best changed-block information available, skips blocks the archive already holds, and sends the rest to SHA inside the encrypted appliance connection. EBA stores the result as deduplicated, compressed recovery-point blocks, shared globally within the repository.

Backup data path

Data the archive already holds is skipped at the source, so even a full read does not mean a full upload.

Backup dataChange information (supported CloudStack KVM backends)

How Changed Data Is Identified

VMware

vSphere changed-block tracking. The first backup is full; later backups read only blocks changed since the previous successful backup.

CloudStack KVM + snagent

For supported configured backends (QCOW NFS, ZFS, LINSTOR DRBD), snagent supplies an extent map so EBA-ST reads and sends only the identified extents.

Sendense CBT

Used for RAW Block and whenever snagent is absent or not applicable. Where no change information is available, EBA-ST reads the full disk while skipping unallocated regions where the platform exposes them.

Transport Outcomes

Both outcomes come from the same SNA-side smart transport role; the difference is how changed data is identified.

Host-assisted CBT

EBA-ST is using an extent map from snagent for a supported configured backend (QCOW NFS, ZFS, or LINSTOR DRBD).

Sendense CBT

EBA-ST is using the Sendense CBT path because snagent is absent, unavailable, not applicable, not selected, or the configured backend is RAW Block.

Source-Side Deduplication

Before uploading, the backup data path checks which blocks the EBA repository already holds and skips sending them; only lightweight references are recorded for those blocks. This applies on every transport path, so even a full read does not mean a full upload.

Deduplication is global within a repository, so identical data across different VMs and recovery points is stored once.

Interrupted Backups

Sendense tracks backup progress against per-disk change state, so an interrupted backup can resume from the last consistent point instead of starting over. Where a clean resume is not possible, the next backup re-establishes a safe baseline.