Sto
Software-defined block storage for high-performance applications
Software-defined block storage for high-performance applications
Software-defined
Sto is a software module that installs on all storage servers and uses shared NVMe devices connected via NVMe fabric across the entire data center.
Block storage
Sto provides fault-tolerant logical volumes with data protection and predictable IOPS/bandwidth/latency for applications.
Logical volumes can be presented as local block device or delivered via NVMe/TCP to any host OS without any additional software.
Logical volumes can be used by any local or parallel filesystem and database.
Logical volumes can be presented as local block device or delivered via NVMe/TCP to any host OS without any additional software.
Logical volumes can be used by any local or parallel filesystem and database.
Composable
Disaggregated
Disaggregated
Sto it abstracts shared CPU and NVMe resources within cluster, aggregates them into pools and uses it on demand. Compute, storage and network resources can be increased as needed.
Scalable
Multi-active
Multi-active
Sto scales from a single server to a 16 node cluster. Each node is truly active, which guarantees a linear performance scaling.
High-performance
Sto delivers best-in-class per-node performance due to proprietary algorithms.
Lockless datapath and data services guarantee the fastest encoding/decoding checksum calculations, imperceptible thin provisioning, minimal disruption in a degraded state, multi-tenant QoS and the most efficient CPU utilization per IOPS.
Lockless datapath and data services guarantee the fastest encoding/decoding checksum calculations, imperceptible thin provisioning, minimal disruption in a degraded state, multi-tenant QoS and the most efficient CPU utilization per IOPS.
Multi-workload
Sto delivers unprecedented performance for any data-intensive workload. It is verified using latest industry-standard benchmarks and modern applications.
Hardware-independent
Sto supports servers, enclosures and NVMe devices, available on the market. You can use any NVMe capable storage hardware, which fits your performance, capacity and cost requirements.
Cost-effective
Sto reduces storage TCO through the most efficient data center resource utilization without proprietary hardware lock-in.
Best-in-class per-node performance
Each storage setup with 24 NVMe SSDs
25 000 000 IOPS
Random Read 4KiB
100 GiB/s
Bandwidth
150 usec
Latency
Sto is developed for the new era of NVMe-based software-defined data centers and provides solutions for HPC, Enterprise, Cloud, Media & Entertainment.
Business Value
Lowest TCO & highest performance
Performance leadership
Forget about tens and hundreds of inefficient storage resources and useless infrastructure overgrowth
1
Hardware-independent
Freedom in designing cost-effective storage infrastructure meets your performance, capacity and cost requirements
2
One fabric
One high-performance network infrastructure across the entire data center reduces design complexity, deployment and maintenance costs
3
Multi-workload
One storage setup guarantees predictable performance and reduces application TCO
4
Cost-effective solution
Per-node performance
Single-node setup delivers up to x5 performance than other solutions.
Recommended configuration:
Linux OS, AMD processor, PCI Express 4.0, NVMe SSDs, 200 Gbit/s network.
Single-node setup delivers up to x5 performance than other solutions.
Recommended configuration:
Linux OS, AMD processor, PCI Express 4.0, NVMe SSDs, 200 Gbit/s network.
Sto reduces storage TCO through the most efficient data center resource utilization.
Per-cluster performance
You only need 4-node setup to achive 100 million IOPS, while others need up to 20 nodes.
Recommended configuration:
Linux OS, AMD processors, PCI Express 4.0, JBOF, 200 Gbit/s network.
Per-cluster performance
You only need 4-node setup to achive 100 million IOPS, while others need up to 20 nodes.
Recommended configuration:
Linux OS, AMD processors, PCI Express 4.0, JBOF, 200 Gbit/s network.
One storage - all workloads
Consolidate applications with diverse workloads and performance requirements across multiple volumes.
Sto optimized for modern workloads and data-intensive tasks: Parallel filesystems, Databases, Data Analytics, Containers & Virtualization.
Sto optimized for modern workloads and data-intensive tasks: Parallel filesystems, Databases, Data Analytics, Containers & Virtualization.
Software Architecture
Sto is a software module that installs on all storage servers and uses shared NVMe devices connected via NVMe fabric across the entire data center.
Sto follows Composable Disaggregated Infrastructure (CDI) concept - a data center architectural concept where compute, storage and network resources are provided as services.
Sto follows Composable Disaggregated Infrastructure (CDI) concept - a data center architectural concept where compute, storage and network resources are provided as services.
Sto consists of two components - core engine and management platform.
sto-core abstracts shared CPU and NVMe resources within cluster, aggregates them into pools and enables to create fault-tolerant logical volumes with data services including data protection and thin provisioning.
sto-platform provides centralized, intelligent management and monitoring within cluster.
Logical volumes can be presented as local block device or delivered via NVMe/TCP to any host OS without any additional software and can be used by any local or parallel filesystem and database.
sto-core abstracts shared CPU and NVMe resources within cluster, aggregates them into pools and enables to create fault-tolerant logical volumes with data services including data protection and thin provisioning.
sto-platform provides centralized, intelligent management and monitoring within cluster.
Logical volumes can be presented as local block device or delivered via NVMe/TCP to any host OS without any additional software and can be used by any local or parallel filesystem and database.
High-performance algorithms
Lockless datapath & data services
Erasure Coding
The fastest encoding/decoding checksum calculations
Thin volumes
Imperceptible thin provisioning
Multi-active
Truly active-optimized nodes guarantee a linear performance scaling
CPU friendly
The most efficient CPU utilization per IOPS
Healthy degraded
Minimal disruption in a degraded mode
Data services without disruption
Rebuild or scaling processes have no performance impact
Composable Disaggregated Infrastructure
Data-centric architecture with efficient resource provisioning
Predictable performance
Enjoy guaranteed performance rate with linear scaling
NVMe over Fabric
Enjoy NVMe/TCP technology and use high-performance network infrastructure across the entire data center
Multi-tenant QoS
Set target IOPS/bandwidth for high priority apps or performance limit for low priority apps
Scale
Compute, storage and network resources can be increased as needed
Learn more about Sto!