Storage software vendor FalconStor Software Inc. last week moved to improve the fault tolerance of its products by acquiring software maker IP Metrics Software Inc. and revealing plans to add server failover capability to its own IPStor product.
FalconStor will add IP Metrics’ NIC Express to its lineup of data pooling software. NIC Express creates redundant data paths between network servers and workstations, and reroutes network traffic, ensuring business continuity if a network adapter fails.
FalconStor paid US$2.5 million cash for IP Metrics.
FalconStor’s IPStor product combines network-attached and storage-area network (SAN) data into a common pool of information that can be shared by users and managed from one interface. It consists of software installed on a server that is positioned in the data path between network servers and the storage devices.
Because the IPStor server handles all the data flowing between servers and storage devices, it is critical that it be mirrored with another server for redundancy, analysts say.
“Putting a single point of failure into a big network is entirely illogical,” says Steve Duplessie, senior analyst for Enterprise Storage Group Inc. “The success that companies such as FalconStor and Datacore have had is pretty impressive, but the smart implementations always have at least two appliances for fault tolerance.
To alleviate this concern, FalconStor has added mirroring, replication and snapshot back-up capability to its products. A function called SerVivor, which FalconStor will release this fall, lets a standby server take over the operations of any of the primary servers if they fail.
With SerVivor, an agent is installed on the server designated as the standby server. The IPStor server monitors the other servers and if one fails, it signals the standby server to take over the failed server’s storage. The standby server effectively takes over the workload of the failed machine, ensuring constant uptime.
Steve Kedem, senior network engineer with law firm Weil, Gotshal & Manges in New York, uses IPStor to connect Hitachi, Compaq and MTI disk drives to Compaq servers. He has created a redundant network with Brocade switches, dual host bus adapters and IPStor servers.
“We have dual redundant paths to each IPStor server,” Kedem says. “If a server can’t get to the storage because an IPStor server failed, then no matter how redundant the rest of the system is, we have no assurance we could get to data.”
Kedem also has tested SerVivor.
“If we have four identical servers each with different operating systems – they could be NetWare, Windows NT, Linux or Solaris – the IPStor servers will listen to the servers and if they realize that Server A has failed, they will tell Server B to boot itself from the SAN,” Kedem says. “We can effectively cluster across different operating systems and applications and protect our data.”
FalconStor declined to comment on the price or availability of the SerVivor option.