ParaScale Architecture Overview

The ParaScale Cloud Storage (PCS) software aggregates disk storage on multiple standard Linux servers to present one or more logical namespaces, files are accessible via 100% standard file-access protocols (NFS, HTTP, FTP). Applications and clients don't have to be modified or recompiled to use PCS, and a single administrator can manage hundreds of servers. Deployments can start at several terabytes and scale to petabytes within a single pool. ParaScale was built to address the needs of emerging markets while being able to scale to hundreds of nodes in a loosely coupled architecture similar to the google file system. We do not require custom or dedicated hardware and can leverage existing IP networking interconnections.

The ParaScale architecture is unique in several aspects. We run in user space which provides many advantages including the familiarity of a standard Linux environment, freedom of hardware selection, co-residency with other applications, simplified installation and management and removal of custom kernel requirements.

Freedom of Hardware

ParaScale understands that hardware requirements will vary based on application, user and corporate guidelines. There is no one size fits all and ParaScale will not dictate configurations or markup commodity hardware. Instead we publish basic guidelines and let the user make decisions. For example, if your application is not driven by performance you can select a very modest server and connect a large amount of SATA storage to reduce the overall hardware cost per gigabyte. For applications that require more performance, provision a faster system. Even legacy servers can be added to a cloud rather than discarded. The bottom line is you are free to choose the appropriate hardware characteristics, vendor, configuration and price to meet your storage requirements.

Simplified Installation and Management

Clustered NAS is painful to install and requires tweaks at the kernel level, custom hardware and overpriced appliance packaging. ParaScale is changing this with its PCS architecture. Users can download software from ParaScale, install a RPM image on a server and be off and running. Storage nodes are automatically discovered by the control node and the configuration is automated. Once installed, an administrator can easily manage 100’s of nodes due to the self healing and managing aspects of the cloud. When it comes time to retire or upgrade a server the system will automatically move the data to other nodes to enable a clean and simple infrastructure change. Adding, upgrading or removing storage nodes is an online operation and does not require a service outage window.

Architecture Diagram

Like the Google File System, the ParaScale PCS employs an asymmetric architecture. The Control Node tells Storage Nodes where to put files and manages the system metadata, keeping track of the location of files and replicas, as well as file-locking, file versioning, and access control criteria. No file contents pass through the Control Node.

Alternative products distribute metadata across a clustered system. Distributed metadata incurs increased file-access latency and high cache coherency network traffic, both of which limit scalability and performance.

Other Resources to Consider