Stole this from ntap thread..... Immediatly sent a copy to my wife, an engineer and in 5 minutes I got a response after her, and some other engineers took a quick look......My wife said " dont buy emc or ntap "....In fact, her boss knows of BlueArc ( dont want to mention the name of her company, but it is one of the largest and most well known companies in the WORLD.....several divisions....)and said they are actually one of the companies beta testing it overseas in France, supposedly. Didnt get any specifics, only that supposedly it is the real deal and will change the storage sector immensely......
Anyway, it's certainly food for thought and worth looking into further......I love NTAP bigtime, and have been waiting for my $14 order to fill, but I think I may just pull the order.....supposedly the companies beta testing right now are trying to actually buy the actual systems they are currently testing.....I dont see how this bodes well for emc or ntap....and as a bunch of engineers said at my wifes work "it is ridiculous how simple this really is and far superior to current methods"...Now I am no techno geek by anymeans...dont understand any of that crap...but my wife is one sharp cracker, as is anyone in her division.......they have to be to work on such highly classified and highly important work.....and I tell ya, they seemed to understand the concept and my wife said take ntap off my buy list after her boss said he was actually familiar with it already and they are one of the beta testers supposedly......Mind you...this is all heresay and talk...nothing confirmed or substantiated, but still food for thought for ntap and emc longs, as well as us wannabee longs.......anybody else know of this or have any comments??
DS and thread, comments welcomed on the following....
community.metamarkets.com
BlueArc
In last month's Gilder Technology Report, George Gilder engaged in a little hyperbole. This is nothing new. George is a master of enthusiasm. So much so that most readers now discount his claims a bit. So when the sidebar of his last issue said "BlueArc imperils all the software based network storage appliances, whether it be from NetApp, EMC, HP, Procom, or Compaq" you'll forgive me if I didn't immediately panic.
But then I saw the BlueArc presentation today, and had a chance to pick the brains of the, well, brains of the outfit, Geoff Barall, the CTO. Geoff is an immensely charming, good looking, congenial and well spoken British engineer, a collection of oxymorons. He would never put it this way, but after an hour with him, this is what I come away with:
NTAP and EMC are screwed.
That pretty much sums it up. Gilder has got this one right.
It really seems so simple. Geoff saw a pretty basic problem with the way storage is handled in large applications right now. On one side of the storage equation, you had the fiberchannel, connecting high speed disk drives together. They all talk at 1.7 Gigabits a second. On the other side you have gigabit ethernet connecting state of the art server farms and the like. (Gigabit ethernet actually has a throughput of 2 Gigabits per second, 1 in each direction, if you want to compare it to fiber channel). In the middle you have servers from NTAP, EMC, CPQ and others moving the data around.
But they max out at about 400 Megabits per second. Bottleneck central. And for the geeks among us, it turns out that the big bottleneck is memory. See, all these servers are basically just UNIX or Windows based PC's, which use regular old PC architecture to shunt data between the storage side of the server to the network side of the server. Grab some data, stick it in memory. Take it out of memory, spit it out on the network. They also use memory to run the operating system and all the control software. Big bottleneck.
Geoff just said "hey, what if you just designed hardware from the ground up that was just designed to move data around."
And then he built it. He and the 80 odd engineers at BlueArc. But rather than get into the hardware biz in a big way (printing their own chips somewhere in Taiwan), they opted to use field programmable chips from Altera. They're cheap, and their upgradable. They waste them. Need to talk TCP/IP? Grab a chip and program TCP/IP into it. Need to talk Fiberchannel to the diskdrive? Grab another chip. Need to speak to an NT file system? You get the point. At every point in the process of moving data where an action is required, you'll find a chip.
The end result? A file server that is 5 to 10 times faster than the fastest competitive machine. A machine that can take 100 times as many user connections, and handle 30 times as much storage, all because it's designed to do one thing, and one thing only. It's a basic principle of engineering - specialized systems are more efficient than generalized ones (at a cost in flexibility).
And of course, BlueArc is matching Network Appliance price point for price point, making the IT decision as easy as possible.
The guy next to me in the presentation leaned over and said "so basically they're selling crack to CTOs?"
That's it exactly. It seems almost too good to be true. It's one of those stories where you say "yeah, but does this actually work"? Geoff says product ships in the next 60 days, so we'll see soon enough. Their beta clients are saying they want to buy the units they're currently testing.
And what can we, the lowly stock investor, do? Nothing. The company is private. We can avoid buying NTAP and EMC I suppose. We can look to Altera as a beneficiary (or Veritas perhaps, with whom they have compatability). We could look at Emulex on a whim, which makes the fiberchannel hubs they expect most clients to use (super high end systems will be delivered with Brocade hubs, Geoff grudgingly admitted, with not-good-for-Brocade-fans body language).
But the sizzle is definitely in BlueArc, and right now, I think the important thing is just not to bet against them. It's so simple really. But no one else is doing it, at least not yet, and BlueArc has a hefty patent portfolio.
And Geoff is going to be a very rich man.
---
Dave Nadig
MetaMarkets.com
=============================================
More on BlueArc: Message 15491966 |
