Friday, June 24, 2005

London OpenSolaris User Group

So we had the first LOSUG meeting Monday night.

(See some other comments by: Gary, Sean, Peter Harvey, Chris Beal, and Chris Gerhard.)

Overall, I enjoyed the evening, although we overran rather badly - as a speaker I have to shoulder some of the blame for that, of course. And I'm looking forward to the nxt one!

Thursday, June 23, 2005

What's your disk MTBF?

A typical disk has a quoted MTBF of a million hours or so. (For SCSI; IDE may be somewhat less.) Let's call that 100 years - allowing for a mix of IDE and SCSI drives.

Now I have about 1,000 drives. Of a wide range of types and ages - IDE, SCSI, FC-AL. Some dating from the late 1990s. Based on the MTBF and the number of drives, I would expect to see 10 disk failures a year - or almost one each month.

I'm not actually seeing anything like that failure rate. It's much lower. Had a disk fail earlier in the week, but that's rare. I'm guessing that I'm seeing only a third of the expected rate of failure.

That's on my main systems anyway. Those are in a controlled environment - stable power, A/C provides constant temperature (not as cool as I would like, but stable). They spin for years without being provoked. I have a 12-disk D1000 array that sat unused in a cupboard for the best part of a year before being rescued and thrown together for beta testing ZFS, and so far 4 of the 12 disks in that have failed in the last year. Which is a rate much larger than you would expect based on the rated MTBF.

My conclusion would be - and this should be well known to everyone anyway - that if you take care of your kit and give it a nice stable environment in which to operate, it will reward you with excellent reliability. Beat it up and treat it like rubbish, and it will refuse to play nice.

Yikes - had to log out!

Finally had to throw in the towel and log out of my workstation (a SunBlade 2000 with 2 gig of memory) this morning.

There's obviously a memory leak somewhere. I had Xsun up at 2G, mozilla up at 2G, several gnome items (especially the panel) were at half a gig. Restarting the clients didn't free any memory in the X server, and the whole box was starting to grind.

So I had to log out to get Xsun to restart. Mind you, the machine had been up (and still is) since Feb 2nd - and I had been logged in that long. That's when Solaris 10 came out, so it's clear that stability isn't an issue.

Thinking back over the last few years, that's probably the longest I've been logged in continuously. Simply because, as a beta tester, I would upgrade or completely rebuild the box every month or two.

And, unless the first update to Solaris 10 arrives real soon, the next time I log out is likely to be on my last day as I turn out the lights.

Wednesday, June 15, 2005

Some Modified OpenSolaris Utilities

One of the things that you can do now that OpenSolaris has been released is to modify the way the Solaris utilities work to suit your own preferences.

So I've done just that and created some modified versions of utilities like ptime, prtpicl, du, and the ucb ls and df.

These just scratch a few of my itches and aren't terribly earth shattering. But they show that it's amazingly easy to modify the OpenSolaris source and gain value from it.


Tuesday, June 14, 2005

Somebody just lit the blue touch paper

Stand well back! OpenSolaris just got let loose!

I've been involved in the pilot for months and it's been a long road, but it's so exciting to finally get here.

Now, off to that source code.


Monday, June 13, 2005

Moved my Solaris Zone

OK, so in preparation for the place I work being closed down, I've moved my Solaris Zone to a new home.

This includes our Postfix under SMF page, and my Java Kstat project (recently updated!).

Friday, June 10, 2005

London OpenSolaris User Group

As Simon Phipps and Stephen Harpster have already noted, we have the first ever OpenSolaris User Group meeting in the UK - organised by Ulf Andreasson - at Sun's London City office at 6pm on June 20.

This will be great. It's good to finally be able to join in the OpenSolaris meeting buzz, to meet our local CAB member, to catch up with some old friends, to get to know more people from the OpenSolaris pilot, and to meet more people interested in Solaris and OpenSolaris.

We haven't confirmed all the speakers yet, but I hope to be talking about my experience with the Solaris development process, from the Solaris 10 platinum beta, through the OpenSolaris pilot, and beyond.

I'm looking forward to meeting lots of you!

Technorati: and

V250 EOL?

Looking at the Sun V250, I've just noticed that it says No Longer Orderable.

What's going on? The V250 is actually one of Sun's better machines. It's the only tower server they had in their lineup, and the only machine with a decent amount of internal disk capacity.

So basically Sun are only interested in customers with expensive datacenters with exclusively rack servers and separate storage arrays. And I haven't seen anything to indicate that they're going to introduce anything else into their lineup.

It's hardly surprising that Sun are struggling to grow sales if they keep reducing the market segments their hardware addresses.

Thursday, June 09, 2005

Solaris, OpenSolaris - which to run?

With the imminent release of OpenSolaris, we have the prospect of one or more OpenSolaris-based distributions such as SchilliX as an alternative to Solaris proper.

Would I use such a distribution? Almost certainly not. Not that I have anything against an OpenSolaris-based distribution, but my primary interest in OpenSolaris is that it's the base for Solaris proper. As such, the "distribution" I'm most likely to run is something along the lines of Solaris Express.

So if I'm happy with Solaris Express, what benefit do I get from OpenSolaris? Well, quite a lot actually.

Simply being able to look at the source is an invaluable aid to understanding the behaviour of the system and debugging problems as they arise. Looking at the source, it can be immediately obvious what the problem is. (Or, equally, that the problem isn't where you first thought so you can eliminate that line of enquiry.) Something that has often frustrated me in the past is that I've often been able to make a fairly confident guess as to the nature of a problem, but haven't been able to confirm it. With source code, I can trivially check; without, I have to wonder whether to fight through the process of a service call. (Note that Sun should also benefit from its customers being able to do problem diagnosis themselves.)

As a true open source project, you won't need a service contract or have to log a support call to report a bug. If something isn't right, you'll be able to report it without any hassle. This is something I've wanted - even as a contract customer - for years. The overhead of raising a support call is sufficient that many smaller problems I spot would go unreported. And the overall quality of Solaris will only be improved if bugs get logged. So making it easier to report bugs should lead to a general increase in the quality of Solaris.

Beyond the ability to report a bug or ask for an enhancement is the opportunity to fix the problem myself. Either for my own personal use, or to be put back into the main product. I'm not going to be writing kernel modules or filesystems or device drivers, but there are plenty of irritations in the standard commands that could be addressed.

My primary interest in all this is Solaris. Sun have decided that OpenSolaris is the development mechanism for Solaris going forward. As such, getting involved in OpenSolaris is a natural thing for me to do. Other distributions look to be a lot of fun, but (by their very nature as being a distinct distribution) aren't going to do the same things that Solaris proper does. So while I'm going to watch them with interest, I'm going to be actually running Solaris or Solaris Express.

Monday, June 06, 2005

Step by tiny step

In TuxJornal, Theo de Raadt says:

"The list of new developments is impressive, but in my view not nearly as impressive as the small little details that continue to be fixed during each development cycle.

Development of OpenBSD is not a milestone-driven series of revolutions. It is a series of small evolutionary steps headed which continue to become cleaner, tiny step by tiny step."

This is interesting, because it highlights the weaknesses I've seen in Solaris development. You see, the problem I have with Solaris is that it's great at the big things, but leaves minor imperfections untouched. Zones and Dtrace are stunning features, and the general solidity and architectural integrity of Solaris is awesome. But Sun have historically spent little time and effort on fixing the tiny little flaws or irritations that are too small (for them) to bother about.

There is hope. With OpenSolaris just around the corner, we have the opportunity to scratch those itches and get to fixing all the little things that have bugged us ourselves. I have my own list of pet peeves ready and waiting to be fixed.

Technorati: -

Saturday, June 04, 2005

StorageTek Buy - ???

So Sun buys StorageTek.

Huh? I just don't get this.

There's not an awful lot of overlap in product terms. Sun already resell StorageTek kit to fill in the gaps in their storage line. (And if you take the 3rd party stuff out, Sun's product line is full of gaps.) So I don't see much consolidation of product and hence cost reductions.

Also, I can't see how StorageTek kit is going to fare being sold into environments containing other vendors' kit. As I see it, an independent vendor is going to be more credible than one owned by a competitor, so StorageTek/Sun is going to sell less kit into other environments and, while it may sell more kit into Sun environments the results is a net loss.

I don't see any organisational benefits from the merger, and I can't see it helping sales.

So why buy StorageTek at all? Why shell out the cash? (And it's a lot of cash!) The Fujitsu (SPARC chip) and Hitachi (high end storage) deals solved a number of problems without dipping deep into their pockets.

Personally, if I had all that cash burning a hole in my pocket I would have gone after some of the smaller high-tech companies that are developing emerging technologies. In areas like InfiniBand or next generation ethernet, for example. Or innovative system builders like they did with the Kealia acquisition.

But having said all that, Sun's track record in making a success of the assets it's acquired isn't good. Time will tell if the Kealia bet works out (so far, it looks good but is taking its time to mature), but I can't think of any others that haven't sunk without trace.