Stuff
Sun cracks me up. They really are the classic example of "it's not what you say; its how you say it". HP used to get the rap as the worst PR/marketing company, but that baton has been passed. Forget about the Netapp lawsuit stuff and let's just focus on this week. First, Sun announces they are rolling their Storage group back under the Systems group. I didn't get one single call on it. Not one. That means no one cares, or people simply assume things. The fact is that it is a good move. Sun has always been a systems company, was forced to spin out a Storage group in order to attempt to compete, learned what that world looks like, and then is rolling it back up to leverage all the good engineering talent/tools, etc. in the Systems group. Second fact is that there is no such thing as "storage" anymore – everything is storage, and everything is a system. Yes, we are going back in time again, but that's the way it is. The lines of demarcation are really blurred with virtualization technologies, and will get fuzzier. A storage device is a specialized application engine containing processors, memory, and physical storage. Sounds like a system to me.
Anyhow, to prove my original point, Sun makes this move and thinks it's big news, but no one cares. They didn't get any accolades, because A: no one cares and B: no one gets accolades for re-orgs. Having said that, this is weird – they released their SPC numbers on the 9900V (big HDS system) – which are stellar – and what system were they run on? That's right, IBM. If you want the world to believe your integration story, you might consider using it yourselves. Yikes. There is much work to be done in Sunville, but I think the rollup can only help, and certainly couldn't hurt.
Speaking of IBM, they made a great decision a few years ago by concentrating on their own server base to reclaim storage share, since their server base was in growth mode. Sun needs to do the same. Plugging into IBM AIX boxes is nice, but if that is your target market, it's time to pack it in.
A few of our guys just came back from an IBM analyst event and said the choreography was superb. I guess they really did a great job of making their SMB blade system really show up HP's variant on everything from noise to power to capacity expansion to ease of use.
Check out what IP.com is doing. People don't realize that intellectual property is critical, and their own employees don't know what is IP vs. what isn't. Why take the chance? IP.com helps folks make sure that intellectual property is protected and defensible. Pretty good idea.
Another thing of complete brilliance is the Data Robotics Drobo personal RAID array (sort of). For $500 bucks you can buy a 4 slot self-contained, completely moron proof system that takes any SATA disk – any size capacity – mixed sizes – and automatically creates RAID like redundancy no matter what disk goes where or which is ripped out and upgraded, etc. It pools all of the total capacity, automatically stripes and places data and parity-like recovery components around, and then changes itself when you remove or add new capacity. If I started with 2x500GB disks for example, and wanted to add capacity, I plop in a 750GB disk and do absolutely nothing else. The system reconfigures itself in the background, moves things around, and now my useable capacity goes from 500GB to 1TB (the biggest disk is always nuked as the redundancy capacity total). Later I add a 1TB disk and now I have 1750GB free to use. Since I'm full on slots, when I want more I just rip out a 500GB disk and plop in another 1TB disk, and so on. So if this $500 box can do all that, why can't my $500,000 dollar high-end array?
Finally, forget the VMware hype – how about the road warrior need for virtualization? RingCube's MojoPac lets me keep my desktop – everything but the O.S. – on a portable drive (or an iPOD!) so all I need to do is plug it into any machine running XP and that machine becomes my desktop. It's secure (encrypted) so when I lose it, no one else can use it, and if my laptop craps out I just plug the MojoPac into any other XP machine. We're in the process of standardizing everyone's desktop, and lord does this make it easy. The stuff is cheap and has worked perfectly. I can't figure out why you wouldn't want to use it, it really is brilliant.
I've got to go make sure the IT community on the great island of Aruba is all OK, so I'll be busy for the next week, and then I'll see you at the Pats vs. Cowboys game, also known as SNW.



I would also like to put my 2 cents in for the drobo. This little box has made my life so much easier - no more raid, no more tech issues. Just throw in another drive (or a bigger one) when it gets full, and you're back in business.
Posted by: danaldonova | October 05, 2007 at 10:27 AM
I think virtualization is the wrong way to go. It's a good temp fix to a more persistent problem. Throwing all your resources into developing more virtualization products is death. It's a fad that will pass. I know it sounds kind of dorky, but I'd like to see Sun come out with a unbeatable line of laser printers. That's infinitely more useful to me, than virtualization, mostly because of the legal requirements for record keeping and somewhat because of my position of not trusting anything, especially anything electronic, when it comes to storing data(bad experiences).
Thanks for the link to the test sheet.
----Thou doth miss the point a bit, me thinks. The laser printer IS virtualization. 20 years ago you couldn't even plug a printer in to your network without getting a PhD first. Though all the exact same engine, each OEM made it so that only thier version would work on thier systems. Want a DEC LPS20 (?), it's $45,000 and needs DecNET to function. Want the exact same printer from IBM? It was $110,000 and only worked on Token Ring. It's virtualization - the abstraction of the silly - that let's you buy whatever printer you want and plop it onto your network at will. The point of virtualization is to keep raising the demarcation line of abstraction so that all the horrific (but necessary) manual labor tasks/functions below that line can be done automatically and invisibly. It's not a panacea, but it's absolutely a necessesity. It exists everywhere today, but we take it for granted and then fret about the next implementation. I think we're better of pushing it faster than slowing it down, but maybe that's just me. Good thought - Steve
Posted by: some dood | October 14, 2007 at 03:18 PM