Data on-prem. GPUs in the cloud. Bridge them at line rate.
Burst to the cloud when the compute is there, pull results home when the job is done — file or object, across major providers. zx saturates the link both ways and skips the 30–50% penalty that drags down standard cloud paths. It's a fast, network-based AWS Snowball alternative — move data over the wire instead of shipping disks.
Hybrid cloud is easy to draw and hard to move.
- Your data sits on-prem while the GPUs and elastic compute you pay for sit in the cloud.
- Ingest and egress are too slow to burst on demand. Deadlines slip and rented clusters idle, waiting on bytes.
- File-versus-object mismatch between tiers turns every move into a custom translation project.
- Cross-provider moves crawl and rack up cost, locking you into whichever cloud your data landed in.
One mover for every tier, every direction.
No impedance mismatch
Move file↔object natively. On-prem filesystems and cloud object stores talk directly — no custom translation layer, no staging detour.
Go where the compute is
Stage data to the cloud the moment the GPUs are there. Pull results back as soon as the job finishes, and cut the time rented clusters sit idle.
On-prem, private, public
One mover spans on-prem, private cloud, and major public clouds — so cross-provider moves stop being slow and one-way.
Wire speed both ways
Drive line rate, scaling out with no software ceiling. A co-designed, DPU-embeddable data path bypasses the 30–50% cloud-stack penalty — inbound and outbound alike.
Zettar moved an actual petabyte over a 5,000-mile network loop in 29 hours — with encryption and checksumming — at 96% bandwidth utilization.

Common questions
What is hybrid-cloud data movement?
Moving data between on-premises, hybrid, and multiple clouds so compute can run wherever it is cheapest or fastest.
How does Zettar avoid the cloud-path penalty?
Standard cloud transfer paths lose 30–50% to stack overhead. zx is engineered to bypass that and run at wire speed.
Does it scale as our data and sites grow?
Yes — zx is symmetric and scale-out: add servers and aggregate throughput grows, with no software ceiling, across as many sites and clouds as you run. Your environment sets the limit, not the data mover.
How much faster is Zettar than cloud-native tools?
In a real KEK-to-AWS transfer (1.2 TiB over 10,851 km), zx was 5.9× faster than aws-cli — about 40 minutes versus 235.
Does distance hurt cloud transfer performance?
Very little — zx is largely latency-insensitive, validated on 10–100 ms paths and over 5,000- and 12,375-mile transfers.
Which clouds and storage does it work with?
Major public clouds and S3-compatible object storage, plus on-premises — file and object, in both directions.
Does it lock us into a cloud or specific hardware?
No. zx moves data in native file and object formats across major clouds and S3-compatible storage — no proprietary wrappers — and the appliance runs on HPE ProLiant enterprise servers.
Do we need to install anything in the cloud?
No. Nothing from Zettar runs in the cloud. The zx Appliance stays on your premises, and zx Object acts as an S3-compatible client, reading from and writing to cloud object storage directly.
Put your data where the work is.
Burst to the cloud, pull results home, move across providers — at line rate, file or object, with the cloud penalty engineered out. See it move your data.
