While storage solutions will always require that hardware be involved, software-defined storage allows the ability to provide storage services without being tied to any specific hardware.
ViPR works by separating the control plane from the data plane. This means ViPR come in two parts, ViPR Controller, and ViPR Services. ViPR 2.0 makes additions on both sides.
ViPR 1.0 offered object-based storage services and HDFS. ViPR 2.0 expands upon these and adds block-based storage services.
For object, ViPR 2.0 supports Amazon S3, Atmos, and adds Centera CAS API support.
HDFS services in 2.0 remain much the same.
ViPR 2.0 adds block-based storage services. These services are enabled by ScaleIO and by the addition of OpenStack support. Initially, at least, these block services are iSCSI only.
The HP SL4540 can be configured with:
Interestingly, HP lists the form factor of a fully-configured SL4540 as 4.3U, which is the first time I can recall having seen a non-integer number of Rack Units listed for a piece of hardware.
ViPR 2.0 adds enhanced support for both VPLEX and RecoverPoint. This support allows ViPR to work across more than one physical data center. Data can be distributed across the sites.
Additionally, data can be replicated across sites under ViPR’s control. This allows for improved data protection and reliability. ViPR will also replicate indexes and metadata across sites.
In 1.0, ViPR provided native support for:
Version 2.0 adds support for:
In addition to supporting more arrays natively, ViPR 2.0 allows for auto-discovery of supported arrays. It also now allows the addition of pre-provisioned environments.
ViPR 2.0 supports the use of even more arrays than those listed above through OpenStack.
Third-party arrays supported through OpenStack include:
While arrays on the “native support” list can be auto-discovered and automatically added to ViPR Controller, adding an array supported via OpenStack is a manual process.
ViPR 2.0 adds additional data protection options, both locally and across sites. Administrators can define failure domains at multiple levels: individual disk, node, rack, and data center.
Additionally, rather than a traditional RAID configuration, ViPR uses a patent-pending “hybrid encoding” mechanism that enables data protection with very low storage overhead.
ViPR 2.0 is planned for GA before the end of Q2.