Hyper-Converged Virtual SAN Tech Specs

Technical Specifications for DataCore™  Hyper-converged Virtual SAN software

Product Details  
Vendor DataCore Software Corporation
Product Type Hyper-converged Virtual SAN
Market Space Small, Medium and Large IT environments along with Cloud Storage Infrastructures
Product DataCore™ Hyper-converged Virtual SAN
Description DataCore™ Hyper-Converged Virtual SAN software (HVSAN) delivers a highly available and high performance solution for virtualized applications, including databases, in a cost-effective package. The software runs directly on hosts to aggregate their individual local spinning disks and flash storage into a fast and highly-available, virtual storage pool. This enables you to web-scale and share data across the cluster of servers without an external SAN. The complete set of advanced features is available for VMware and Microsoft environments.
Announced January 31, 2011 [new licensing model and rebranding in April, 2014]
Current version / Release date 10 PSP5, June, 2016
Updates Available for download for customers under annual maintenance contract
Packaging Supplied as a downloadable CD image with context sensitive Help files
Licensed By Storage capacity under management, per node, some separately-priced optional features

Supported Environments  
Main Features Device-independent synchronous mirroring, asynchronous remote replication, continuous data protection (CDP), online snapshots / backups,  high-speed caching, random write accelerator, auto-tiering, QoS, storage pooling, non-disruptive disk migration, thin provisioning,  deduplication , cloud integration, NAS/SAN unified storage , centralized management, analysis and reporting
Host Connectivity Fibre Channel (up to 32 Gbps), iSCSI (up to 40 Gbps), and Fibre Channel over Ethernet (FCoE) via connection to FCoE switch 
Access Type Block disk I/O over a physical or virtual SAN. File system access is provided via NFS/SMB (CIFS) protocols from the underlying Windows Server operating system. The two access methods may be combined to meet high availability, unified storage (SAN/NAS) requirements.
Host Environments Supported Virtual machines on computer systems, physical desktops and virtual desktops running standard Windows operating systems including (Windows Server 2012, 2012 R2, 2008, 2008 R2 2003, 2000, Windows 8, 7, XP), UNIX, HP-UX, Sun Solaris, IBM AIX, RedHat Linux, Suse Linux. As well as virtual machines hosted under VMware vSphere ESXi, Microsoft Hyper-V, Citrix XenServer, Linux KVM and other server hypervisors.
Storage Supported (back-end) Any internal drives, external drives, external disk arrays, JBODs, Solid State Disks (SSD), Flash memories, and intelligent storage systems supported on Windows Server 2008, 2008 R2, 2012 & 2012 R2 may be attached to the DataCore node(s). They may be internal, direct-attached and SAN-connected. Cloud storage is supported via Cloud Gateways.
Storage Interfaces Supported (back-end) Any disk interfaces supported by Windows Server 2008, 2008 R2, 2012, 2012R2 including SAS, SATA, iSCSI, Fibre Channel.
SAN Switches Supported All standard iSCSI and Fibre Channel switches are supported
Network Interfaces Standard IP network interfaces for internode communications, console access, and asynchronous remote replication between nodes. Standard IP LAN connections for file sharing (NAS).

Operating Platform HVSAN is installed on standard Windows Server 2012, 2012 R2 or 2008 R2 platforms. These may be physical or virtual servers.
Operating System Requirements Windows Server 2012, Windows Server 2012 R2, or Windows Server 2008 R2 SP1 with .Net 4.5.1
Processor Standard Intel/AMD x86-64 CPUs
Memory Required (min) 8 GB RAM
Cache Up to 8 TB of processor memory (RAM) per node may be used as high-speed cache
Storage Capacity Model dependent; up to 1 Petabytes per node
Max Nodes in a centrally managed group 64 nodes per group -  Several groups may be managed from a central DataCore console.
Hypervisors Supported HVSAN may be deployed on any of the popular server hypervisors including VMware ESXi, Microsoft Hyper-V and Linux KVM. It runs in a Windows Server guest VM.

For Hyper-V, HVSAN runs in the root partition and does not require a guest VM.

It can also run natively in a bare metal Windows Server (such as one providing Failover Clustering for a specific application – e.g., SQL Server)

IOPS/ node Dependent on configuration of underlying hardware platform. (see SPC-1 results)
Bandwidth/node Dependent on configuration of underlying hardware platform.
Max # of Virtual Disks There could be a limitation on number of Virtual Volumes created on each HVSAN node, but the limitation will be defined by size of each Virtual Disk and performance requirements per HVSAN node.
Min size (GB) of virtual volume 0.001 GB
Max size (GB) of virtual volume 1 Petabyte = 1024 TBs
Max # of Physical Volumes The max number of Physical Volumes is dependent on the Storage sub-system hardware
Physical Volumes (max size) 1 Petabyte
Host Initiators (max #) A single HVSAN node may contain multiple (depending on number of slots on the motherboard) multi-port Fibre Channel (FC) HBAs or iSCSI NICs that would be connected to a number of other nodes in the cluster. 

Snapshots / Backups (Differential & Full clones)

HVSAN supports instant point-in-time snapshots along with full copy clones. You may periodically update snapshots to a later point in time with just the changed blocks that transpired after the last snapshot. You may also use snapshots to restore the source volume via the source update option. Snapshots are readable and writeable. HVSAN uses copy-on-first-write technology as well as thin provisioning to significantly reduce the space taken up by incremental snapshots. The snapshot is available immediately upon request and may be triggered by VSS-compatible applications, VMware vCenter and PowerShell scripts. DataCore works with popular 3rd party backup products including CommVault and Veeam to simplify end-to-end data protection and ensure rapid data recovery at a granular level.

Continuous Data Protection (CDP)

Continuous data protection may be selected for virtual disks to restore the state of the virtual disks before an undesirable change (ransomware, user errors and virus, database corruption, for example). The software will log and timestamp every update to the virtual disks in a rolling buffer covering up to the last 14 days (depending on configuration parameters and available disk space). The state of the virtual disks may later be recovered to a desired point in time within the 14 day period.

RAID Levels

HVSAN supports software mirroring and striping across physical disks behind each DataCore node. It can also rely on RAID protection from back-end RAID subsystems and disk controllers.

High Availability

DataCore offers an enterprise-class “zero downtime, zero touch” high availability storage solution that prevents equipment and site outages from interrupting critical information flow. The cost-disruptive software innovations proven in over 30,000 global deployments provide simultaneous access to active copies mirrored between physically-separate locations, making it the centerpiece for business continuity.

Synchronous Mirroring

DataCore enables application and/or server clustering technologies in campus and metro-wide configurations (stretch clusters) to continuously update and retrieve data, even with the loss of an entire site. In contrast with competing alternatives, no manual intervention or scripting is needed for failover, resynchronization or failback; everything is automated. Customers report a 100% reduction in storage-related planned and unplanned downtime; many for more than 8 years. The software synchronously mirrors virtual disks in lock step across two DataCore nodes. Separations up to 100 kilometers (~60 miles) are typically achieved using dedicated high-bandwidth optical routes across Metropolitan Area Networks (MANs). Longer distances may be achieved for more latency-tolerant use cases. Diversely routed redundant mirror connections between nodes ensure no single point of failure.

Asynchronous Remote Replication

HVSAN supports Asynchronous Remote Replication over conventional LANs and WANs using standard TCP/IP protocols. It automatically compresses transmissions to reduce bandwidth requirements. Secure, encrypted connections such as VPNs and trunked or aggregated multi-link circuits can be used to enhance the privacy and speed of inter-site transmissions. Disaster recovery sites may be initialized via network connections or by cloning virtual disks to transportable media at the primary site and shipping them to the remote destination. Replication is bi-directional. Topologies supported include 1-to-many, many-to-1 and many-to-many. (Note: An individual virtual disk may not be replicated to more than one remote node)

Multi-pathing Support (Linux)

HVSAN supports multi path drivers for different Linux versions and different OS types.

Multi-pathing Support (Windows)

DataCore MPIO supports Auto failover and Auto Failback between primary and alternate paths. It also supports failover and failback between Fibre Channel and iSCSI inside a single host.

Multi-pathing Support (VMware)

HVSAN supports different versions of VMware ESXi using different path selection policies (ALUA), including round robin and most-recently-used.

Load Balancing

Explicit load balancing is achieved by spreading virtual disks across multiple ports. Multiple back-end channels to LUNs within a device are supported. If one back-end channel fails then all LUN I/O activity will continue on the remaining healthy back-end channels. Additional load balancing occurs across disks of the same tier in a pool. HVSAN periodically looks for any devices that have unusually high activity (hot spots) compared to the other members of the pool. It then automatically fine tunes the pool by moving disk blocks to less active drives.

Non-disruptive upgrades

Non-stop data accessibility is configurable with paired redundant HVSAN nodes that circumvent any single point of failure. Such innovative and cost-sensitive architecture ensures that service level obligations continue to be met in spite of component failures. Essentially, the I/O responsibilities associated with an outage, scheduled or unexpected, are distributed in real-time among the remaining storage resources being managed by HVSAN. Nodes automatically exploit alternate paths to the other running HVSAN node with the mirrored copy to maintain end-to-end high availability. When taking a HVSAN node down for a prolonged period, control of that node’s shareable back-end storage resources can be transferred to a standby node without disrupting applications. The standby node takes the place of the one undergoing maintenance to maintain active mirrored copies of data for high availability. In addition, the standby node ensures that the overall throughput and responsiveness of the system remains intact.

Online Disk Expansion & Removal

Disks can be added to a HVSAN node to expand the configuration on line without any down time. Also, when using thin provisioned disks, if more storage is needed, you can increase the pool size, without affecting any host or any Virtual disks presented to those hosts. Disks may also be removed non-disruptively from a pool. Simply select the drive to be taken out of service then, in the background, HVSAN migrates the blocks across the remaining drives in the pool. Once complete, the physical drive may be removed.

LUN Security / Masking

HVSAN does not use LUN Masking technology. All FC or iSCSI ports connected to a node are discovered automatically. Once an FC port WWN or iSCSI IQN are discovered, and a node is registered, a Virtual disk may employ that port. Only those nodes in the same cluster that have been granted explicit access to a virtual disk can detect that the disk exists. Other hosts will not see it.

Virtual / Physical Correlation (Troubleshooting)

Within HVSAN, each Virtual disk may be created from multiple physical resources (storage). HVSAN provides detailed information about which physical pool the Virtual disk is created from and which ports are used to present the Virtual disks to the hosts.

Automated Storage Tiering

Sub-LUN migration: Dynamically promotes groups of most frequently accessed disk blocks (chunks) to fastest class of storage and demotes infrequently accessed chunks to lowest cost, slower tiers. Administrator can set tier preferences and override migration for special circumstances. Up to 15 tiers can be defined to distinguish between devices having different price/performance/capacity characteristics. Multiple devices are supported in each tier.

Thin Provisioning

Takes the guess work out of how much disk space to allocate to an application. Configure up to 1 PB virtual disks and HVSAN takes care of allocating actual disk space as the demand for physical space increases from

Deduplication and Compression

Post processing deduplication and compression may be selectively enabled for specific virtual disk pools. These processes reduce the space necessary to store data by eliminating duplicate blocks and compressing the single image, when possible.

High-Speed Caching

HVSAN assigns configurable amounts of the memory (RAM) inside each server that it has been installed on (DataCore node) into a high-speed storage cache. Advanced caching techniques inherent in HVSAN's design accelerate the response time of concurrent reads and writes from multiple application servers to virtual disks on the storage area network. The performance enhancements come inexpensively, exploiting the low-cost memory of the commercial processor platforms on which HVSAN runs. HVSAN's cache is closely analogous to that found in modern high-end storage subsystems. 

It resides between the Hypervisor and the physical storage. Like caches found on storage subsystems it provides a variety of caching services noted below:

  • Read-ahead: When a request for a block is satisfied, HVSAN will automatically pre-fetch adjacent blocks into its cache on the principal that if "block X" is required, "blocks X+1 and X+2" will probably be requested shortly afterwards.
  • Write-behind: Unless specifically told not to cache writes, HVSAN will respond with "I/O complete" when a request is cached and mirrored to another node (multi-cast stable storage). It will flush (destage) the cached request to disk when convenient.
  • Write-coalescing: One of the reasons that writes are not normally flushed immediately to the disk is to allow HVSAN to better organize the sequence of writes and to concatenate writes from adjacent blocks into a single operation while the physical disks are busy.

HVSAN's caching acceleration applies to all manufacturers' storage devices configured throughout a Storage Area Network. The caching strategies implemented by HVSAN have been thoroughly tested and proven effective on generation after generation of hardware.

Random Write Accelerator

Increased performance for workloads characterized by many random writes, such as frequently updated databases, ERP and OLTP systems, are possible by selecting the sequential storage option (random write accelerator) for virtual disks.  The technique uses additional capacity to expedite the data transfers. 

Quality of Service (QoS)

QoS parameters can be set for nodes or groups of nodes as well as groups of virtual disks. For streaming applications which burst data, it’s best to regulate the data transfer rate (MB/sec) to minimize their impact. For transaction-oriented applications (OLTP), limiting the IOPS makes most sense. Both parameters may be used simultaneously.


HVSAN commands may be issued programmatically rather than from the administrative console using an extensive library of PowerShell scripting commandlets. Essentially, any command available from the DataCore Console GUI may be scripted as well.

Alerts and notifications

E-mail alerts and event logs provide clear notifications when the software detects unusual conditions including failure prediction notifications from S.M.A.R.T.–enabled storage. SNMP v1 queries and traps are also supported for interfacing to popular Systems Management and Systems Monitoring packages.

Storage Profiles

Set up relative priorities among Virtual disks by classifying their Storage Profile as Critical, High, Normal, Low, or Archive. HVSAN functions like auto-tiering, remote replication and synchronous mirror recovery will give preferential treatment to higher priority volumes over less critical ones when resource contention exists. The storage profiles may be customized.

Private, Hybrid & Public Cloud

Extend the value of private, hybrid and public clouds through Software-defined Storage services provided by HVSAN software. Access public cloud storage through Cloud Gateways and integrate directly with the OpenStack Cloud Operating System through Cinder block storage services.

Ready to see how DataCore can take your enterprise IT to the next level?
Request a personalized demo today!

Get a Demo