Effectively, the only difference between iSCSI and FC performance is with bandwidth-intensive applications. The iSCSI protocol is extremely efficient and all of our testing has shown this to be true. ESG Lab, which is our hands-on testing service, recently did an apples-to-apples performance test with the EMC AX100i (iSCSI) and the AX100 (FC). The lab analyzed different e-mail workloads, and we simulated an e-mail environment. We used an industry standard benchmark tool, IOMETER to run the workload. The maximum I/O rate recorded by IOMETER was translated into the maximum number of typical Exchange users. Our calculations were based on guidelines from Microsoft. The AX100 with FC supported up to 424 e-mail users and the AX100i with iSCSI supported up to 399. This is a marginal difference.
iSCSI should work with any application that FC does. Every customer must perform their own evaluations to determine if there is a fit and which vendor they want to use. I have spoken to dozens of iSCSI customers and they all were satisfied. Some even stated that they would never go back to FC.
We have seen very little deployment if iSCSI HBAs in application servers. Approximately 90% of deployments that we are aware of are not using iSCSI HBAs. Typically, the customers that have installed iSCSI HBAs are using them for remote booting functionality rather than performance, because the processing power in today's servers is more than sufficient to handle the load.
This was first published in May 2005