High performance network storage systems.
Stackscale offers two different network storage flavours: All-flash that guarantees stable maximum performance and very low latencies and Hybrid, featuring a great balance between performance, guarantees and price, and thus achieving an outstanding quality/price ratio.
zScale Flash Premium
This type of storage is designed to provide high performance in IOPS and bandwidth in any circumstance with very low latencies. In other words, it is designed so that the storage never becomes the bottleneck of the system.
- The observable performance from a single VM (virtual machine) varies from 6K to 40K IOPS depending on the type of workload, and a bandwidth of between 200 and 400 MB/s using blocks of at least 32KB.
- Average latency of writing/reading below 1.5ms.
- Performance guaranteed by SLA with penalties.
- Ideal for all types of projects. Especially for databases and high performance requirements in reading and writing.
zScale Hybrid storage combines RAM, SAS SSD disks and rotating SAS disks to provide the best quality/price ratio possible in environments that despite requiring 100% availability, do not require constant and guaranteed very high performance, and can afford typical rotating when requesting data that is accessed infrequently, commonly called cold data.
Thanks to its extensive RAM cache and high performance SSD, the system provides the great majority of operations latencies and IOPS levels characteristic of All-Flash systems and is only limited in terms of available bandwidth to ensure performance to all customers.
The observable performance from a single VM varies from 4K to 35K IOPS depending on the type of workload, and a bandwidth of between 100 and 150 MB/s.
The zScale Flash Premium systems are strongly oversized to ensure a spectacular performance in all circumstances, avoiding the effect of noisy neighbours.
Stackscale deploys its own network storage systems based on highly available dual active/active controllers and dual-domain SAS environments that are fully fault-tolerant with no single point of failure. The systems are based on the proven and well-known enterprise-grade ZFS filesystem.
Shared specifications among the storage flavours are as follows:
- Highly available network-attached storage (NAS) delivered through the NFS protocol, which provides a better data integrity protection than other block-based technologies (iSCSI, FC, …) under certain contingency scenarios.
- Bottleneck-free storage access by using very high capacity 20Gbps links at each computing node.
- Storage efficiency technologies: Thin provisioning and Inline compression. These technologies are normally present in most centralized storage, but the profits remain in the hands of service providers. In order to promote transparency, Stackscale transferred these benefits to its customers, where each contracted 1TB can easily become effectively 3TB and more.
- Non-disruptive maintenance of the platform.
- Backups based on snapshots included, by default configured to take a snapshot every 12 hours, retaining a maximum of 4. Customizable upon client request.
- Full replication included, synchronized hourly and including backups, fully independent system located in a different location within the same geographic area. Protects information against disasters affecting the exceptionally active storage system and/or the physical environment. Possibility of replication to geographically distant environments.