Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Son of SAN - Storage Networking Technologies -- Ignore unavailable to you. Want to Upgrade?

To: Douglas Nordgren who wrote (1005)	1/28/1999 9:06:00 AM
From: Sam	Read Replies (1) \| Respond to of 4808

The following was posted on the MTI thread by someone who is, I think, a reseller and integrator of computer systems (David Lethe, some of you may have run into him on one or more of these threads). I am reposting it here asking if anyone has any comments on it, in particular the last paragraph regarding "the back end", but any other comments which clarify, agree, disagree are welcome to this non-engineer. Seems to suggest that SCSI and FC can coexist without any problem. ================================================================= To all: Let me explain the true meaning, value, and future for FC technology. There is a lot of incorrect information on this thread, and I hope I can help all understand. First, FC is undoubtedly the future for at least the next few years, as a host interconnect. That means it is absolutely going to be the defacto standard for host-to-subsystem interfaces. This is because it not only offers a very high pipe, but promises to become a standard of which all CPU and disk manufacturers will be able to support. (Note I said PROMISES. We aren't there yet. FC as it is today is still in it's infancy, and by no means does everybody's adapter work with everyones hubs, software, switches, or disk subsystems. In fact, the ANSI standards still haven't been defined for this. As SANs (Storage Area Networks) become a reality, FC will be the clear interface of choice. FC promises to allow topologies which allow hundreds or thousands of computers, regardless of operating sytems, to connect to hubs and high-speed switches, along with disk subsystems of any vendors. Devices are appearing today which also allow legacy SCSI-based (the type of FC that MTIC and Most vendors use is actually a flavor of SCSI, specificly serial SCSI, instead of parallel SCSI) systems to attach to a fibre backbone. Fibre can also be glass or copper, or a combination. After you do the physical connections, then you need software which must run on every system to actually share data. Most vendors, including EMC have something in the works, and some vendors have software with limited O/S support. Expect lots of announcements and positioning now, and many products to ship before year end. Some UNIX's allow two systems to share without adding anything. FC is actually the means to the end. The true benefit of FC is that for the first time, an interconnect mechanism is possible which can allow multiple computers, REGARDLESS of O/S, to share data, at backbone speeds instead of TCP/IP (networking) speeds. Real-world, this promises a 100X throughput advantage when compared to standard 10Mbit ethernet. (That is why 3COM is getting into SAN software. SANs will put a big hurt on ethernet and ATM, which is what people are using today for data sharing). On the back-end, however, FC is not only unnecessary, but it is undesired. An intelligent disk subsystem with a FC host interface, and disks that are either UltraSCSI or SSA is what you want. The big reason is cost. Implementing fibre at the disk level is significantly more expensive and complicated. Until disk drives start pushing data much faster, there is no technical benefit, albeit some marketing benefits. For example, what good does an 8-lane highway do you if there is only one car on the road. Also, FC as a disk interface can actually be significantly slower than SCSI with small block I/O, which is the norm for most business I/O (transaction processing). These are the reasons that the giants such as EMC haven't implemented fibre-to-fibre solutions. EMC has (arguably) all the money-in-the-world for R&D, and they don't have such a product. Neither does IBM, or Compaq's (DECs) storageworks. Happy Investing -- David