SGI hopes to emerge as a leaner, meaner organisation by the end of the third quarter. The hardware maker this week filed an amended reorganisation plan that calls for it to finish off bankruptcy proceedings by September. If all goes as expected, SGI will trim its total debt down to USD 70m from USD 345m. Some of the debt will be removed in exchange for the privilege of investing more money in SGI 2.0.
SGI 2.0, oh God, I cannot believe you just said that.
Atleast it isn’t:
“SGI Next Generation! Leaner, meaner, and even a cuter logo!”
Or “Its continuing mission… to boldly go where no workstation co. has gone before!”
SGI’s been notorious for huge and expensive marketing ploys in the past, hopefully they’ve mended their ways. They have good products, but that doesn’t really mean anything when you’re spending every penny you’ve got to keep up a flashy image. They also underwent a major clamour over Itanium that seemed to have been spread like the plague by HP-related execs, which also kicked SGI in the butt. It shows how quickly bad management can drag down a market leader with lots of technical talent. Good luck, SGI.
They have good products
Really? Like what? Everything they do can be done cheaper by a Linux cluster these days, it seems.
I think SGI finally realized that clusters are eating away at their market. So they decided to come out with their own
http://www.sgi.com/products/servers/altix/xe/
One biggest problem with white-box linux clusters x86 is hardware failure. If they use high quality components like they used to with their workstations then this helps solve that problem.
If they can offer a powerful cluster at a decent price with a good support contract they might have a winner on their hands
Ironically, some of the first successful Unix clusters were built using SGI workstations, at Fermilab, in the early and mid 1990’s.
well yes as the other reply says sg finally realised that their market was shrinking because of clusters, however cheaper does not equate with better. Altix has pretty much lowest latency, highest bandwidth, supersize IO, largest memory on the market, largest linux server you can buy, ….
> xxmf wrote:
> well yes as the other reply says sg finally realised that their market was shrinking because of clusters, however cheaper does not equate with better. Altix has pretty much lowest latency, highest bandwidth, supersize IO, largest memory on the market, largest linux server you can buy, ….
Two problems with what you’re saying:
– It’s incorrect. Lowest latency & highest bandwidth, supersize IO, largest memory – none of it.
– And not including the above comment : Who are the buyers for the largest Linux server with the largest memory on the market with supersize IO, lowest latency and highest bandwidth? SGI has primarily been kept alive by a) sale of storage, b) a few corporations who were unable to make the shift in time and c) national labs in various countries.
The problem is, even though SGI 2.0 offloads $280M of debt, they as much as ever need to show profit and growth. They see storage as a cash cow (which it may very well be, outside of jurisdiction, we only stick to Lustre on regular Linux clusters) but I just can’t see SGI only be a storage company.
“- It’s incorrect. Lowest latency & highest bandwidth, supersize IO, largest memory – none of it. ”
OK, who else built a 13TB machine?
For the other. What is wrong with storage sales, when the storage sales are in fact driven by the IO architecture?
What is wrong with selling to national labs for that matter…
but to answer your question: yes they have rather niche/boutique customer list and need to break out into the commercial enterprise space. The sad fact is altix is a good engine from that, but those people are very conservative – they switch supplier rarely and certainly not for a co. in the financial mess sgi got itself into.
Yes I didn’t respond about whether sgi will do well post-11, and they can’t be a pure storage play I agree. What I understand is that part of the problem that is preventing them from geting into a profitable scenario is the ongoing costs associated with that 280M debt, as well as a bunch of other burdens associated with once being a Big Company^tm. The 11 gives them the wiggle room to fix some issues, obviously they have to be doing better on an ongoing basis too, but lowering the bar doesn’t hurt at all.
I hope they make it. I’ve watched them sadly ever since I first heard of them( they were on the way down even then). I’ve always been very impressed with their products, and couldn’t, for a long time understand how they could be so technologically impressive and sooo brain-dead managerially at the same time.
Is SGI still SGI or should it be S(!G)I. Didn’t they sell graphics department?
SGI had great hardware and their Irix OS.
Ohh, i would love to see that OS ported to x86/x86_64 hardware.
Irix is one of the most stable OSes i have ever used.
I can’t agree with that, sure they have very good hardware but Irix is painful.
When I was administering Origins, I was much more busy to make them work well than now with another unix brand.
I remember that some daemons like nis were quite unstable, fam was eating all cpu. Those servers required much attention, and needed to reboot every month to keep in good shape.
Also the sales people were very unpleasant.
SGI had great hardware and their Irix OS.
The key word here is ‘had.’ Now, it’s horribly outdated – the last major release was 6.5 in 1998, and they haven’t had a major version number change since the early 1990s.
I realize that numbers aren’t everything, but eight years stuck on 6.5.x is a sign that IRIX either reached the pinnacle of operating system technology in the late 90s and there hasn’t been anything that could possibly want improvement since then, or that it’s been stagnating and has fallen behind the times. A quick comparison of the experience of using my FreeBSD box with my Indigo2 workstation leads me to think it’s the latter.
sgi are commodoty now and they cant compete with dell. sgi 2.0? shesh. no thanks.
Last I checked, SGI was never really going for the same market as Dell – sure they may have some overlap (Dell’s mid-to-high range storage _might_ compete with SGI’s storage products), but I don’t think there’s really all that much to compare between the two. SGI has gone after very high-end systems as of late, and Dell is certainly in the commodity market space.
I agree – sgi 2.0 is a dumb name
“One biggest problem with white-box linux clusters x86 is hardware failure.”
You do realize that one of the points and advantages of a cluster is that individual nodes can fail without it affecting the system as a whole, right?
Personally, SGI should:
1) Embrace Itanium and bring down the costs of their workstations and servers which use it – bring it down to the price level where by their price/performance is on the mark with AMD/Intel x86 offerings; either via raw price/performance, or simply offering more products and services with each sale.
2) Embrace Solaris, and port it to Itanium, licence the Sun compilers, and get them optimised for Itanium – rename Solaris, IRIX version 7 to demonstrate a clear and clean break from the 6.x series in reference to the architecture switch and operating system overhaul.
3) Either do one of two things, buy out Qt (relicence under something like CDDL or BSDL) and embrace KDE as their default desktop, or muck in with Sun and improve GNOME; make management tools easy to use, so that those who aren’t necessarily technically computer literate (scientists) can setup and manage a cluser of computers without the need of having an expensive IT department.
4) Stop using the excuse ‘we’re going for a niche’ – a niche company is a dead company; niches only exist for a small period of time before they end up changing and morphing into something else; SGI need to sell not only big friggin servers to scientists, but sell them to those businesses who want a big 128way machine, loaded to the gills with memory, and Oracle running on top of it.
Edited 2006-07-08 04:15