CAStor is high performance object storage software that combines intelligent automation and Elastic Content Protection to transform commodity servers into easily scalable, fault tolerant storage that can be used to store massive amounts of content for any amount of time.
CAStor is a complete software stack that boots from bare metal and includes all operating system, services and interfaces needed. It runs everything from RAM, including its distributed object index, leaving up to 98% of hard drive capacity for net content. That's why rolling upgrades on CAStor are natural and straightforward.
CAStor doesn't rely on a file system (zeroFS) or database (zeroDB) that both get brittle with size. All operational and descriptive information is encapsulated as metadata with each object. The zeroFS and zeroDB approach means storage capacity and object count can scale quickly.
Use the servers, hard drive technology or network of your choice. Seamlessly add new hardware while mixing node types, hardware vendors or drive sizes, even within the same node. This lets you easily and steadily add new, more efficient hardware while retiring older nodes at the press of a button. Forklift upgrades are a thing of the past.
Darkive is patented adaptive power conservation technology that will selectively spin down drives and power step node CPUs to reduce power consumption by up to 80% in cold archive use cases. These savings put Darkive optimized CAStor clusters well within the TCO of large tape based media archives while still ensuring content is readily accessible.
All CAStor nodes can handle all operations. An innovative algorithm and caching is used to dramatically reduce overhead and manual management associated with other architectures like RING. There are no single points of failure like controller nodes, databases or management nodes.
Object addressing is based off of a 128-bit unique universal identifier or unique name. When content is accessed it is located instantly (zero IOPS) and delivered in milliseconds regardless of stored capacity or object count.
CAStor's zeroFS, zeroDB, Swarm architecture, and support for replication and erasure coding provide consistent rapid performance for small (Bytes) or large (Terabytes) files.
CAStor can extract every bit of value from commodity hardware but if a power boost is needed you can choose higher performance servers, solid-state drives or 10 GigE networks to enable performance intensive workloads.
CAStor's unique Swarm architecture enables the massively parallel interaction between each individual node automating capacity and load balancing. Distributed repairs also happen independently because every object is aware of a lost replica or lost segment and re-replicates accordingly. Swarm even powers Darkive and is key to directing requests to optimal nodes allowing all others to fall asleep and save power.
Lifepoints are administrator-defined policies stored as metadata with the object enforced by the Health Processor. Lifepoints are used to automatically manage the entire content lifecycle from creation to expiration including number of replicas, erasure coding scheme, delete protection and deletion.
CAStor's no single point of failure approach truly shines when it comes time to scale. Simply rack servers, boot CAStor and store. There is no LUN or namespace management, and all storage balancing is automated.
A mind numbingly large namespace combined with Swarm architecture enables massively parallel scaling to hundreds of petabytes and hundreds of billions of objects.
With the parallel nodes very loosely coupled and no bottlenecking controller nodes or metadata databases to hold you back, just add the hardware of your choice for linear scaling of capacity and throughput. You can do this at your own pace, fifty nodes at a time or smoothly one by one: no constraints for cluster expansion whatsoever.
Supports single or multi-site deployments (with massively parallel replication, both synchronous and asynchronous) to support a broad range of use cases and business requirements.
Combines automated management of replication and erasure coding (definable per object) with continuous integrity checks, passive recovery and active recovery – even for erasure coding leading to much higher durability numbers versus competitors with similar protection schemes.
Erasure coding can be used to save footprint and increase data durability while replication can be used to ensure rapid access. CAStor also automatically replicates hot content for reliable content delivery regardless of access patterns.
CAStor is highly available by design, supports hot plug drives, adding/retiring disks/nodes and rolling upgrades of the full software stack – all without service downtime.
Address regulatory mandates that content is stored on non-erasable, non-rewritable media. You can also use Legal Hold to create a point-in-time snapshot of a specified set of objects at a specified time that are then immutably stored regardless of what happens to the original object or cluster.
Gain robust insight into objects stored via (the optional) Caringo Indexer. Ad hoc queries can be run on object attributes or metadata and viewed through a web-based portal or compiled as JSON or XML for import into 3rd party analytics applications.
All customers have instant access to our knowledge base, FAQs, tech notes and product alerts. We also offer higher touch support packages depending on your business requirements.
Get ahead of content growth – optimize your storage resources.
Store massive amounts of content for any amount of time.
The key to a successful, secure cloud storage service.
Software Development Kit with HTTP interface and documented API for Java, Python, C++, and C# for CAStor object storage software.