Scale Computing’s ICS™ Architecture

As storage continues to evolve, a shift is occurring from the traditional scale up approach toward the more flexible scale out strategy.

In scale up, capacity is added by upgrading to the next larger filer level, until the largest is reached. This architecture requires failover clusters to provide high availability and redundancy. It’s both time consuming to manage and relatively inflexible, as growth is dependent on the storage vendor’s smallest upgrade path. For instance, if a filer’s drive capacity is reached and you need an additional TB, the only option is to purchase the next larger system available. In other words, companies are forced into a continuous cycle of over-buying forklift upgrades, and data migration.

Global Namespace

ICS (formerly TrueCluster) is a scale out architecture. It uses software to create a global namespace across all nodes in the cluster, thus creating a single cluster with virtually unlimited scalability.

Higher Throughput

ICS is unique in that it includes both file and block-level services. While data is stored at the file level, it is striped across all nodes in 256K blocks. This highly parallel architecture allows us to deliver enhanced performance by combining the I/O performance of all disks, network interfaces and processors in the cluster.

No Single Point of Failure

Each ICS cluster node contains the information necessary to access all data blocks simultaneously and can assume responsibility for any other node in the entire cluster. If a drive or node fails, services can continue uninterrupted. That’s not the case with controller-based or scale up architectures. Controllers hold the metadata for the filers below. If it fails, access to data stored on the drives is interrupted.

Transparent Fault Tolerance

In scale-up architectures, fault-tolerance is addressed by adding redundant paths to multiple controllers. ICS, however, contains its own block replication system. When a node fails (determined by a quorum of surviving nodes), the other disks prevent it from writing and continuing their operation.

Replication

ICS contains advanced, asynchronous replication features that can be architected in a one-to-many and many-to-one scheme. Synchronous replication is available through Scale’s professional services group.

Unlimited Scalability

Since ICS creates a global namespace, it has virtually unlimited scalability. Systems can scale per TB well into the petabytes.

Future-Proof

Unlike controller-based systems that typically are programmed to read only nodes of a particular density, Scale’s ICS can accept nodes of multiple densities to the cluster. Nodes purchased today will work with nodes purchased in the future.

NAS/SAN Consolidation via a Unified Architecture

Because ICS contains file (CIFS, NFS) and block (iSCSI) level protocols, companies can combine their existing SAN / NAS solutions onto a single storage server. Both NAS and SAN protocols can be served from every node in the cluster simultaneously, while still maintaining centralized management.

Processing and Throughput Scalability

Another benefit ICS brings to storage is its ability to scale both capacity and processing power. Because each node in the cluster contains its own processor, memory and disks, speed increases as the cluster builds. You can fine-tune your cluster to optimize for price or performance using our configurator.

.