Our systems come standard with 25Gb bonded network connectivity to each physical node within our cloud. Unlike other providers, we provide the same network bandwidth speeds to our high-speed or object storage solutions too, so you don't experience any issues when accessing your data. For 90% of our customers, these speeds work incredibly well for their deep learning workflows; however, if higher bandwidth and decreased latency are needed, we have high-speed solutions to meet those needs.
Cirrascale Cloud Services provides its customers with the ability to go beyond the standard and to experience connectivity up to 3200Gb with NVIDIA InfiniBand NDR networking. In fact, we have the ability to connect your dedicated multi-GPU cloud servers together with the industry's fastest, ultra high speed all NVMe Flash storage as well for the fastest available feeding of large deep learning training requiring exceptionally high IOPS per server and centralized access.
Private Networking enables our multi-GPU cloud servers to communicate with other cloud servers in the same data center. Systems can be clustered together for replication, larger job analysis, or more. Additionally, private networked servers can connect to the same storage resources making the delivery and sharing of data across nodes easier than ever.
Cirrascale Cloud Services has built its entire high-availability network around top-of-the-line providers in our state-of-the-art West, Central, and East data centers. We provide an uptime SLA around network, power and server availability. If we fail to hit our target, just call us and we'll credit you based on the amount of time your servers were unavailable.
Ready to take advantage of our flat-rate monthly billing, no ingress/egress data fees, and fast multi-tiered storage?
Get Started