Slashdot Mirror

It was too busy "competing" with MS to focus on it's core market - high performance computers.

Sun did waste a lot of effort on desktop initiatives. (Still does.) But that's never dominated management's attention. It's just what got the most press, because the issues of hardware manufacturing and sales don't make good copy.

(Right now, I'm working on a system that's in the process of being upgraded from Hypertransport 1 to Hypertransport 3. Doesn't that send chills up and down your spine? No?)

The problem is not that management didn't pay attention to the hardware business — they paid plenty. The problem is that they kept selling to the 1998 marketplace long after the game changed. In 1998, there was so much demand for computers, and people were so unpicky about costs, Sun could sell expensive systems just by boasting how powerful they were. Then the dotcom bubble burst, and people either went out of business or survived by looking for ways to do business cheaper. And a big way to cut costs is to switch from proprietary architectures (SPARC, MIPS, PowerPC) to the commodified x86.

Took Sun a long time to come to terms with that. When the bubble burst, the party line at Sun was that it was a temporary downturn and they could just ride it out. Well, the downtown was indeed temporary, but the customers never came back. They wanted commodity systems, and Sun was only working on SPARC systems. Yes, they had acquired Cobalt, but the SPARC-uber-alles mindset at Sun soon drove the Cobalt people away and destroyed a product line Sun had spent $2 billion acquiring. When they finally admitted to themselves that they had to change with the marketplace (and there are lots of Sun people who still haven't drunk that koolaid) they had to build up the business all over again, partly by outsourcing design, partly by buying up yet another x86 company. Ironically, that company was founded by Sun co-founder Andy Bectholsheim, who had left Sun partly because of this very issue.

So, despite the attention it got in the press, the Sun-MS feud was just a sideshow. What really hurt was their inability to adapt.

One of many reasons right here

I messed around with Ext4 for a little while on my machine (Like a couple days, just toying with it and seeing how its performance compares to Ext3 and Reiser4) a while back, like maybe a little bit before it was merged as experimental in the mainstream kernel. It is fast, backwards-compatible and extremely featureful. XFS is not a bad filesystem, but it has some problems, in my eyes. Metadata-only journaling, aggressive caching that makes it a potentially dangerous choice if you don't have a UPS, very slow metadata and deletion operations.

That's great that XFS has a lot of features Ext4 is bringing to the playing field, and has had them for a long time. To pretend, however, that the developers of Ext4 simply have a NIH syndrome is just silly and disregards the fact that there is a lot that Ext4 already provides that XFS doesn't, and even more that it will soon. You might not see what the big deal is, but really, I can assure you that it won't be very long before the new ideas Ext4 employs are in widespread use.

Here's an interesting article that really caught my eye with this: "Storage snapshot: The financial firm has more than 14 Petabytes of active storage and plans to add "several more Pbytes" within the next 12 months."

Steinbach Credit Union is my home bank, so it was interesting to hear how they set this up back in 2003.

SCU has a second branch located in Winnipeg. Data is constantly synchronized between the two sites providing a physical disaster recovery solution and a convenience for customers, as loan information, etc is always up-to-date so it doesn't matter which branch you visit. (People from Steinbach often visit Winnipeg for shopping and movies). As opposed to paying $70,000 per month for 3rd party leased lines, or $1 million to lay their own fibre, SCU found the cost-effective solution to create their own private wireless network. SCU also uses the direct link for email, VoIP, and streaming security cameras which provides additional bandwidth and long-distance savings.

The towers are full-duplex and shoot a narrow microwave beam which is almost impossible to intercept 100 feet above ground and data is encrypted "2^48 power" and apparently not affected by the weather. <<insert Canadian weather joke>>

SCU won the silver medal in the SearchStorage.com Spring 2003 Storage Innovator awards competition for their innovative wireless SAN setup.

Here are more article links with details and diagrams of the setup and equipment used.

Haven't heard much from Scotty lately....

Do they need a place to crash

You are quite correct, there is no requirement for SCSI drives in an iSCSI implementation, iSCSI refers the protocol, not the drive interface, i.e. it's the SCSI command protocol implemented over TCP/IP. So yes, you can build an iSCSI system out of commodity parts and many people are doing so. if you want get an idea of the options out there for doing this, take a look at: http://www.byteandswitch.com/document.asp?doc_id=9 6342&WT.svl=spipemag2_1

More on the shakeup.

We're seeing this more now (think Sun and SGI) -- companies that are underperforming making changes at the top in the hopes of generating new intitiatives and pumping up the stock price. It remains to be seen if all the bloodletting will lead to any marked improvement in the short term -- new execs have to deal with things as they are and try to untangle the mess left on their desk before they can move forward.

Sensitive data leaks, missing plutonium, hard drives with classified data unaccounted for....

Those kinds of failures.

The kind of incompetencies and oversight that can not and should not be tolerated or have excuses made for, whether they are dealing with projects that relate to national security or just studying the fluid dynamics of ketchup. People have been fired for much less in the past.

Google's servers are already pushed to the limit according to this article.http://www.byteandswitch.com/document.asp? doc_id=85804&WT.svl=news1_2 Graphical ads are sure to slow them down even further.

Also, graphical ads are almost certain to be less relevent to the search results than the current text based ads (which is the other reason most people don't mind them). I'm sure mass-market advertisers like Coke or Ford would love to get graphical ads on Google, but how many people are searching for 'cola' or 'crappy+cars'?

Dell, Dell, Dell. The customer at one site decided to buy into Dell's 'home-grown' mid-tier SAN offering in that brief period of time around 2001 after Dell and EMC had parted ways and before Dell came back to its senses and re-partnered with EMC. The re-badged EMC Clariion controllers + arrays on a Brocade fabric had not given us a single issue in the year they'd been in use, but this new demonic half-breed SAN shows up as part of the "new Win2K SAN" (yes, this customer ended up with 4, I kid you not, 4 different and non-connected SANs in the same physical server room).

Dell techs came, and Dell techs went. We had a former field-circus clown who was "certified" on this new storage system sitting in our server room, leafing through the product manual and scratching his head while customers were ranting and raving about not being able to get to their files. The cluster software didn't work. Various bits of the hardware routinely committed seppuku rather than operate with that demon of a storage system. The Dell-trained installers ran the cables backwards between the disk trays and the controller (gee, I wonder where all these fiber-channel errors are coming from). Files mysteriously disappeared. Various VPs within Dell called and made weekly pledges of earnestness in an effort to not get their product thrown out of the server room.

A few months after all this, Dell quietly discontinued their 'home-grown' SAN products and went back to EMC.

I'm happy to use their laptops and desktops as long as someone else pays for it :), and their entry-level to midrange server offerings aren't significantly worse than anyone else's, but may I be damned to the foulest depths of Hell if I ever recommend their storage systems and professional enterprise services to anyone. Ever.

Byte and Switch also reported on the SPECmail news and emphasized CommuniGate Pro's scalability and performance: http://www.byteandswitch.com/document.asp?doc_id=7 9042&WT.svl=wire2_1

More articles, for the more article inclined.

I like the following quote, "3. Stability. Netcraft has announced that Windows 2000 server has finally gone for over 2 years without a reboot." After checking netcraft, we can see their server is at byteandswitch.com. So fellow slashdotters, want to give them a hand?

Let's see if the site is still up after slashdot effect...two years without much traffic may be possible with Windows 2000 but how about under heavy load?

Sun acquired Pirus Networks to help them on a chassis with FibreChannel, iSCSI, and perhaps InfiniBand.
http://www.byteandswitch.com/document.asp?doc_id=2 1423

Before that they picked up Dolphin Interconnect to help them make a 4x (30 Gigabit/sec) InfiniBand Host Channel Adapter.

Here is an article from an EETimes Network site, CommsDesign with some details.

http://www.commsdesign.com/news/tech_beat/OEG20020 919S0076

It is definitly interesting stuff. Everyone is trying to do Shared I/O and I/O Virtualization; maybe Sun can get it right.

Does "OFS" constitute a tacit admission that NTFS wasn't the best thing since sliced bread, but only a retread of DEC's Files-11 filesystem, and that NTFS had all the problems systemic to Files-11, like needing defragmentation?

Or is this just another instance of MSFT using it's monopoly power to screw over people who have bothered to reverse-engineer and implement a "de facto standard" like CIFS and SMB?

I really don't see how any rational being could interpret OFS as any other alternative.

Since MSFT did a half-assed job of copying Mach when they developed NT, they should take a look at Jeff Mogul's doctoral dissertation, "Represeting Information About Files": J. Mogul. Representing information about files. In Proc. 4th International Conference on Distributed Computing Systems, pages 432-439. IEEE, May, 1984.. Maybe MSFT can read this and get it right, instead of half-assing it like 8.3 file names, drive letters, NetBIOS, NTFS, NT, and many other examples.