To: Thomas Mercer-Hursh who wrote (25198 ) 5/23/2000 12:04:00 PM From: buck Respond to of 54805
First, ethernet connections are not just limited by total bandwidth, but also by competition for that bandwidth. When there are a large number of users competing for the bandwidth, the resulting collisions and retries can lower the effective bandwidth quite dramatically over what the pipe could theoretically carry. SCSI, by contrast, is a much more managed protocol and can deliver sustained performance close to rated bandwidth when properly implemented. Is this contrast an issue in this debate? Ethernet switching GREATLY mitigates the collision/retry issue you raise. I would bet that 100% of NTAP filers are implemented on at least one switch port, and the servers that they serve are on switch ports, too. What you don't ask, though, is a more important question today. Overhead on a TCP/IP connection is far greater than on a SCSI bus or using SCSI over a Fibre Channel channel (from the Redundancy Dept. of Redundancy). Etherenet is non-deterministic and connectionless, while SCSI is highly deterministic and connection-oriented. TBL: Connectionless is good at a web transaction level where data transfer size is usually less than 1K, but bad at a data streaming level, where data transfer sizes of 256K are more common.Is there a difference in these architectures between read and write performance? This is going to be true regardless of the flavor of file access, NAS or SAN. And this is where the storage administrators make their dough. It's up to them to make these things zing along as fast as possible. Since the majority of them are trained in connection-oriented, deterministic policies and procedures and practices, I contend that SANs are easier to manage for today's storage admins. There is still a gap in today's enterprises between the system/storage admins and the networks admins, and generally, neither understand the other's domain. This is certain to change, but I frankly have not seen the gap contract to where *I* would like to see it. There is a growing understanding, but the nuts and bolts of each are still foreign. TBL: well, I don't think there is one. I think I might have gotten it right on the first try. buck