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 : EMC How high can it go? -- Ignore unavailable to you. Want to Upgrade?

To: jhg_in_kc who wrote (10274)	5/24/2000 12:18:00 AM
From: buck	Read Replies (1) \| Respond to of 17183

<you see anything that should alarm EMC longs?> Nothing that should keep you from sleeping nights, unless you own a bunch of BRCD, too. <g> Seriously, though, from looking at their management team, it looks like they are very strong on the IP side, and kind of light on the storage side. And IMHO, the storage side is the toughest part. They will be forced to drive the IP processing down into silicon in order to compete with FC throughput and very low overhead. That silicon IP has only been successful to date in very high-end switches. A notable point is that a number of the guys in management came from 3Com...the networking company that launched a SAN initiative, then aborted it...and the same company that built the first programmable ASIC to perform layer 3 switching. So they just might have some helpful experience close at hand, particularly considering 3Com's recent capitulation at the high-end of the network space. Their first, and biggest problem, is writing a driver for an ethernet card that will run storage protocols. No matter what they say, it will inevitably involve encapsulation. The only alternative I can see will be IP headers that do device command and control. More IP headers layered into that 1500 byte packet means more overhead, which means lower throughput. Plus those IP headers will need to go through standards bodies, if they want to run on today's networks, as they are claiming to be their goal. The second problem is having native devices that understand IP, and more importantly, SoIP. It will be interesting to see what comes out of the device cos. like Quantum, Seagate, etc. that will support this model. I know that there are skunkworks projects all over the place that have been trying to do just this, but I've seen nothing close to fruition. Not that that means anything, of course...skunkworks are by nature secretive. The third problem is overcoming the limitation of 1500 byte packets of data. That is just plain inefficient, particularly when compared to FC and it's 2K frame size that can handle very large blocks with one read or write command. IP and ethernet require a lot of host processing to segment and reassemble large blocks of data. Hence, the need to drive IP into the silicon. It's absolutely amazing to me that this announcement of a future technology caused Ashok Kumar to tell everyone in FC to run for the hills. SoIP is decidedly NOT a new concept. It's been talked about for years, simply because everyone loves to see those expensive switches and routers pump even more data over existing infrastructure. But guess what? They built FC because the existing infrastructure can't handle the really high-volume stuff, and likely can't anytime soon. And I don't expect that to change because of one company's really neat announcement, and one analyst's over-reaction. So, IMVHO, I believe that EMC suffers very little exposure from this, far less than that from NTAP. EMC could even profit from it by adopting it. I don't doubt that should this Nishan stuff come to fruition that EMC will almost certainly announce support for it quite quickly. buck PS Has anyone figured out how to pronounce it yet?