coreo.cloud
Telegram Explore servers

03 / HIGH-CAPACITY STORAGE CLUSTERS

United States 10Gbps Storage Servers

United States (Miami) · 10 Gbps

CAPACITY:
PORT:

12 configurations available

LET'S TALK

Storage in Miami to share data across the continent

If you distribute complete datasets to teams or customers in the Americas, a 10 Gbps storage server in the United States shortens downloads. Tell us who reads the data and how often, and we will help you choose.

Message us on Telegram
Server fronts with drive bays and status LEDs
Illustrative photograph · RubinObs/NOIRLab/SLAC/DOE/NSF/AURA · CC BY 4.0
Computing equipment with drives and storage components
Illustrative photograph · Mitch Altman from San Francisco, USA · CC BY-SA 2.0

Plan delivery of complete datasets

The United States Storage 10 Gbps selection offers twelve configurations from 100 to 650 TB, with HDD data space and a 500 GB system SSD. For research datasets or downloadable materials, define the delivery unit: one file, a directory or a selection prepared for each consumer. This affects the volume leaving the server and the space occupied by deliveries being prepared, even when all of them use the same source data.

iperf3 can reverse the direction of a traffic test, providing a reference for sending from the machine with a documented receiver. Then measure a real delivery of the selected dataset and note how long it stays active. If several users request the same material, add that scenario to the planning example. Capacity and machine selection should account for both the underlying collection and the work required to deliver it.

Computing infrastructure facilities and equipment
Illustrative photograph · The National Archives (UK) · CC BY 3.0

Decide how several readers share the data

The twelve United States Storage 10 Gbps capacities use two Xeon E5-2680 v3 processors and 128 GB of RAM. An analysis platform can have readers scanning an entire dataset alongside others needing only a small portion. Classify those tasks before comparing configurations. Useful project concurrency depends on each reader's request and how long it remains active, as well as the number of people using the service.

Observe Linux counters for completed reads and sectors read during a sample containing both tasks. Record each consumer's progress too, so an aggregate figure does not conceal differences between them. Then try a schedule separating full scans from interactive queries. The comparison provides a basis for choosing the material to retain on HDD and the mix of tasks the service should handle within the same working window.

Listed as available
2x Intel Xeon E5-2680 v3
128 GB DDR4 ECC Registered · 100TB HDD Storage + 500GB SSD (OS) · 10 Gbps
700 €
net / month
Configure server