Operating a storage systems in the cloud already comes with enough challenges, why would you also want to take a hit on features and technology? Many available software defined storage solutions miss features such as storage tiering, volume cloning, asynchronous replication, or erasure coding for efficient data protection with fewer data copies.
Simplyblock is different in this regard. With industry-wide expected features, plus the underlying copy-on-write based storage engine, simplyblock provides the most comprehensive software defined block storage solution available. The technology, based on NVMe/TCP and NVMe-of is optimized for I/O-intensive workloads in Kubernetes. Simplyblock runs everywhere, including bare-metal and AWS deployments.
At the core of simplyblock's innovation is our advanced storage virtualization technology (software defined block storage), designed to seamlessly integrate with and enhance cloud environments. Our solution operates as a sophisticated abstraction layer, intelligently managing the complexities of underlying storage systems.
Simplyblock creates a unified storage pool by aggregating multiple storage resources, including NVMes, SSDs, HDDs and object stores. This pool acts as a single, logical storage entity, abstracting away the complexities of individual storage types behind a standardized logical nvme interface. Our distributed algorithm ensures optimal data placement across these resources, balancing performance and cost considerations.
We leverage NVMe over Fabrics (NVMe-oF) technology, specifically NVMe over TCP, to provide high-speed, low-latency access to storage resources. This protocol offers significant performance improvements over traditional iSCSI connections, while maintaining broad compatibility with existing systems. The use of NVMe over TCP allows us to deliver near-local storage performance, even for remote block storage volumes.
Our platform continuously analyzes data access patterns to make intelligent decisions about data placement. Frequently accessed data is kept in high-performance tiers, such as local NVMe storage, while less frequently accessed data is automatically moved to more cost-effective tiers like object storage or HDDs. This process is entirely transparent to the applications, ensuring optimal performance without manual intervention.
Simplyblock implements an efficient copy-on-write mechanism for all write operations. This technology forms the foundation for our instant snapshot and cloning capabilities. When a snapshot is created or a volume is cloned, we only track changes from that point forward, minimizing storage overhead and enabling rapid creation of development or test environments.
Our Container Storage Interface (CSI) driver seamlessly integrates simplyblock's software defined block storage with Kubernetes environments. This driver translates Kubernetes storage requests into simplyblock operations, allowing for dynamic provisioning and management of storage resources. It supports advanced features like volume expansion and snapshot creation directly through Kubernetes APIs.
To ensure data integrity and availability, simplyblock employs erasure coding for its SDS block storage. Similar to RAID-5 but more efficient, our erasure coding distributes data and parity information across multiple storage nodes. This approach provides robust protection against data loss while minimizing storage overhead compared to traditional replication methods.
Our disaster recovery solution utilizes an asynchronous replication mechanism. Changes to data are continuously streamed to a write-ahead log stored in object storage This log can be used to recreate the entire storage state in a different availability zone, enabling near-zero RPO recovery in disaster scenarios.
Simplyblock's architecture is designed as multi-tenant storage. We implement strict isolation between tenants at the storage level, allowing multiple users or applications to share the same underlying storage pool without compromising security or performance. Each tenant's data is logically separated and can be individually managed, encrypted, and optimized.