BHF-Bank, one of Germany's largest financial institutions and part of the Dutch ING Group, has installed StoreAge Networking Technologies Ltd.'s virtualization technology to simplify the management of its storage environment and lower the costs associated with running it (see StoreAge Banks a Customer).
The bank was getting increasingly frustrated with having to spend large amounts of its IT budget on storage management software that locked it into one vendor, namely IBM Corp. (NYSE: IBM), says Stefan Fischer, CEO at Basis Advanced Technologies Ltd, the storage systems integrator hired by BHF-Bank to find a solution to this problem.
BHF-Bank's financial applications are stored in a disaster recovery configuration consisting of two distinct sites, 11 km apart. Each site comprises multiple Sun Microsystems Inc. (Nasdaq: SUNW) SunFire servers, Brocade Communications Systems Inc. (Nasdaq: BRCD) SilkWorm 3800 Enterprise Fabric Switches, and IBM Enterprise Storage Systems (a.k.a. "Shark"). IBM Global Services provides the maintenance for the SAN infrastructure.
The problem, says Fischer, was that the bank didn't want to invest in additional software from IBM for multipath failover, remote mirroring, and snapshot capabilities just for the Shark. So, in stepped StoreAge. "The StoreAge SVM [Storage Virtualization Manager] appliance replaces the full functionality of the Shark," he says, "but its replication and disaster recovery features work regardless of which vendor's RAID system is installed." He adds that the SVM, as well as managing multivendor storage, is also "half the price of the Shark."
Fischer says a key factor in the decision to pick StoreAge over FalconStor Software Inc. (Nasdaq: FALC) or DataCore Software Corp. -- two other vendors shipping virtualization technology today -- was for its asymmetric, or out-of-band, approach to storage virtualization. In a small SAN with just a few Windows NT servers and less than a terabyte of storage, the in-band approach, in which the virtualization appliance sits between the host and the storage device, works fine, Fischer says. "But in a larger configuration, you need a second virtualization engine for failover, in case the first one dies," he says. That would require customers to pay twice the price, but it wouldn't achieve the required performance because of cache coherence issues and bandwidth constraints when transferring all the data from one appliance to the failover system, Fischer says.