The benefits of centralised storage are well known: higher resource utilisation, more control of data and increased ROI to name a few. But with differing transport mechanisms available, how do you choose the right one for your network storage?
Two very popular block-based network storage standards are iSCSI and Fibre Channel. They allow data to be sent via differing protocols – IP (Internet Protocol) and FCP (Fibre Channel Protocol). I’ve described the relative merits of each below.
Network storage and iSCSI
What is iSCSI? A comparatively new and increasingly popular storage networking standard, iSCSI is also a relatively low-cost means to network storage. It uses IP network architecture for SCSI-based traffic.
The two iSCSI options at the host level are software iSCSI and hardware iSCSI. Both provide the same functionality, but with software iSCSI, the processing overhead resides on the host system’s resources as opposed to the iSCSI hardware adapter. Generally speaking, the hardware iSCSI option is the more efficient of the two.
Typically, iSCSI is suited to smaller and less I/O-intense network storage environments, because the overhead inherent in encapsulating iSCSI commands within a TCP/IP frame tends to impact performance. So the greater the number of transactions and amount of traffic on the network, the less the data flow will perform, and the slower the performance will be, between the initiator and target.
Network storage Fibre Channel
Fibre channel (FC) is a mature and robust carrier, widely used when a given application demands high performance and reliability. FC traffic requires you to install specific hardware in the host system (FC host bus adapters) and suitable supporting infrastructure (FC switches/hubs).
Whilst the functional speeds are generally higher, so are the associated costs – both in terms of capital to purchase FC infrastructure and in operational costs. The introduction of new technology into an environment requires specific skills, so additional training may be needed to support this network storage infrastructure.
Making the right network storage decision
As with many technology choices, the most suitable solution comes from weighing the business’ requirements based on price, performance and functionality. Sometimes the decision is simple and the application needs the Fibre Channel “grunt” from its storage network, or the IT service is small enough to dictate iSCSI as the best fit.
Other situations may warrant a mix of solutions. To reduce costs for a company that needs disaster recovery, for example, it might use Fibre Channel networking for local traffic, where performance service-level agreements are demanding, and iSCSI for cross-site replication, where recovery point objectives and recovery time objectives allow.
Allaster Finke is a practice lead at reseller GlassHouse Technologies (UK) and a contributor to SearchVirtualDataCentre.co.UK.