FalconStor ships clustered VTL with de-dupe

By Kevin Komiega

FalconStor Software has released the next generation of its virtual tape library (VTL) product, this time with an eye toward enterprise environments in need of scalable VTLs with global data de-duplication.

The new VTL Enterprise Edition has been integrated with FalconStor’s clustered de-duplication technology and Single Instance Repository (SIR) and is designed with scalable I/O performance and storage capacity to accommodate implementations ranging from remote offices to enterprise data centers.

Scaling VTLs in enterprise data centers can be a dicey proposition, especially when you are trying to get the best performance out of a system’s data de-duplication capabilities. Many users end up with multiple, discrete VTL systems that can produce redundant data and that must be managed individually.

FalconStor’s technology evangelist and spokesperson Diamond Lauffin says VTLs that do not scale can cause more problems than they solve. “You have a head and it can de-duplicate data, but if you can’t scale that head you’re limited to the speed of what that one head can do. That’s not good enough for enterprise environments,” he says. The result, according to Lauffin, are multiple VTL implementations, each creating multiple copies of the same data. “Many vendors quote their de-duplication ratios on a per-head basis, but if you actually need six or eight heads in an enterprise environment you could end up with multiple files based on the number of discrete heads,” he explains.

Lauffin says FalconStor’s VTL Enterprise Edition overcomes issues associated with scaling VTLs through its ability to scale to eight nodes, with each node having a throughput of 1.2GBps. The Enterprise Edition supports sustained backup traffic originated by the backup hosts via Fibre Channel (up to 4Gbps) and iSCSI (up to 10Gbps) at 1.2GBps per node. Cluster eight nodes together and the system is capable of saturating a Fibre Channel or Ethernet backplane.

The VTL can be implemented as a Linux or Solaris appliance or as a virtual appliance under VMware. The architecture enables FalconStor to integrate with third-party backup software vendors to offer features such as NDMP tape copy and a hosted server backup enabler. The VTL can interface with tape libraries via SCSI, Fibre Channel, or ACSLS.

The host backup enabler technology allows third-party backup software to run on the appliance to perform backup, restore, or media copy functions over a PCI bus. The system’s de-duplication architecture supports dedicated N+1 clustering with up to four SIR nodes. The SIR repository can scale up to 256TB of unique data to achieve optimal de-duplication rates.

An eight-node configuration provides 1.6PB of raw storage capacity and can be managed as a single group through group policies and reports. The VTL supports up to 128 virtual libraries, 1,024 virtual drives, and 64,000 virtual cartridges per node.

FalconStor’s VTL Enterprise Edition also supports optional one-to-one and many-to-one off-site replication for disaster recovery. Only unique data blocks are replicated to the remote site, which reduces network traffic by up to 95% based on a 20:1 de-duplication ratio, according to company claims.

The VTL also features a number of tape-related enhancements. Embedded tape security and data encryption, enhanced tape media security with tape shredding, secure tape export, and replication with data encryption have all been included to meet security requirements.

The VTL Enterprise Edition is available as an appliance or as a software-only product through FalconStor’s partners. The company’s OEM partners include vendors such as Brocade, Copan, EMC, H3C, and Sun. Pricing starts at $53,000.

Today’s VTL users face a number of obstacles. According to Laura DuBois, research director for IDC’s storage software practice, scalability, tuning performance and capacity, and replication are all thorns in the collective sides of enterprise users.

“At initial deployment, turning on compression within the VTL is great for space, but it’s a memory and CPU hog on the VTL. So finding the balance is an issue,” says DuBois. “Another issue is ensuring VTL A can replicate to VTL B and, in the event that VTL A goes down, that VTL B can take over so there is cross-system awareness, workload balancing, and fail-over.”

DuBois says to effectively scale VTL deployments, systems need to be able to address higher capacities and scale throughput accordingly.

“FalconStor’s software architecture provides concurrent overlap backups with de-duplication, offering throughput speeds well-suited to scaling to high-capacity enterprise environments,” says Dubois, “and features such as clustering, tape consolidation, and encryption are mandatory in today’s availability-, capacity-, and security-conscious data centers.”

This article was originally published on March 01, 2008