Sponsorship

Monday, April 27, 2009

Aussie Impeller Propels Itself Overseas

I do like this new impeller in this video, it's quite an achievement.  Jay Harman developed it, sounds to me like he might be an Aussie.  His company PAX Scientific seems to be offshore, but I may be wrong.  I wouldn't be surprised though to see PAX overseas, because in Australia we've traditionally discouraged the innovators and forced them to go somewhere else to develop the idea.  Here's another piece of info on the PAX Lily Impellers. It also seems to imply that PAX Sci and Jay Harman are not in Oz any more, Toto...

You have to love that we give the rest of the world so many good minds and killer developments...

Wednesday, April 15, 2009

Head In Clouds, Feet On Ground.

Cloud Computing.  Who hasn't quailed at hearing that term, over and over, for the last year?  It's also not so much the words themselves, as the opinions for and against, all given as Every Absolute Truth You Always Wanted To Know About Cloud Computing But Were Afraid To Ask.  And to me the emphasis seems to be on the "Every" rather than the "Absolute."  No-one really knows how well it's going to work, who's going to be financing and making the clouds, how well Cloudsourcing Company A's cloud is going to interact with Cloudsourcing Company B's cloud, and whether it will be the answer to your prayers or not.

I'm definitely not an expert on any of that.  When I was working I found it was hard enough to keep up with my own server rooms and remote disaster recovery facilities.  I did, however, find time to play with virtualisation and cluster computing.  And improved a few places I worked at, I might add.  I can and will add my voice on the subject of CC.

To begin with, I think CC will benefit small enterprises and home businesses the most.  If you have a LAN and a broadband router going at ADSL2+ speeds, then you will gain a cloudy server room out there, without the cost of owning hardware, the ongoing air conditioning and electricity and consumed space costs.  They'll probably be the first to entrust data and logins to the cloud, the first to use cloudsourced computing power to get their product out to their public.

Home users too will be able to see some gains.  Storage, for a start.  How wikkid to stash your files *here* and then go *there* and be able to work on those same files?  Yeah, totally.  And using a legit copy of Microsoft Office online versus risking being caught with a "liberated" version?  Priceless.  (Or not... %)

But established businesses with more than a few servers will find that it's less percent of gain to them.  Because, you will no doubt prefer to keep local logins, for the time being at least.  And the most sensitive files - that's human nature, after all.  In almost all cases, I see the IT staff will be the ones pushing for this and trying to make a business case for CC.  Management just isn't (sorry CEOs and other ULM) very savvy about IT, because if they were, they'd be in IT instead of management.  It's like saying that because a person is skilled at using surfboards, they are great at making them.  Or vice versa.  People have skill sets.

I had to "sneak" virtual computing into the network, find an old clunker server and slip VMware onto it, then install a server virtual machine and use it for a non-core service, then find another non-critical server service and slip that onto the same machine, until I could reliably switch between the mission-purposed machine and the VM without a hassle.  Then I switched the mission-purposed machines off and waited for one of the random inspections to reveal that we were using less machines and electricity and still fulfilling all our IT needs...

I'd tried to get official sanction for VM technology, but it kept getting put off and put off.  Never outright forbidden mind you, and so I took the initiative.  But I figure CC will need to pull much the same stunt.

There's a few other questions that everyone who wants to use CC will need to answer.  Who's going to pay for this?  Will it cost you a significant outlay and significant ongoing costs, more than half of what you'd expect to pay for keeping these services locally?  Because then, I can see that given the other drawbacks of CC, you might be better off keeping those services locally.

Will there be a significant security risk for you?  How sensitive is the service and associated data for you?  Because, in the case of an attack on a local server cluster after your customer data, for example, your IT department can, as a last resort, physically pull the network cable to that cluster.  I "pulled the plug" on our entire internet connection when the company webserver farm was hit by one of the first ever DDoS attacks back in the early days, saved us a bundle because in those days bandwidth was expensive.

If someone is cracking into your data in a cloudsourced situation, you don't get the option to shut that service down.  Indeed, if a critial set of CC servers goes down you don't even get the choice to call your IT staff and have them do a callout.

Sorry if this is an overly simplistic view of CC - but I'm often an overly simplistic person when it comes to those things, and it's generally stood me in good stead.  I'm often overly conscious of security or reliability implications that my management considered too slim a chance to worry about - and invariably were glad they'd let me plug anyway...  And I can see so many points of failure and breach in CC that it makes my eyes water and my palms sweat.

All that said, I do use Google's mail and apps almost exclusively, with Star Office locally, and I use several other software services "out there" to store and synchronise files, bookmarks, and photographs.  I use services "out there" to collaborate with other people and run projects.  I just make sure I have some form of local backup, is all...

Tuesday, April 14, 2009

Space Tech Comparison

We've come a long way.  I mean, just look at this, which you can find among other things here, and which are to be auctioned off next month.  I remember as a kid playing with components that large, that clumsy, and that simple.

That's a few transistors in those cylindrical shields/heatsinks, seven of them by the looks of it.  A modern CPU chip contains hundreds of thousands of transistors on a single sliver of silicon...

And you think to yourself "OMG!  We sent people into space.  With equipment like THAT..."

Scarily, we now have spacecraft with modern technology and equipment inside, and they don't seem to fly half as well nor fulfill their missions half as often...

Monday, April 6, 2009

Broadband Initiative 2.0

With gardening taking up most of my productive time (and leaving me pretty stuffed afterwards and basically cbf to blog) I've been a bit lax in feeding stories into the blogs. But also in my defense, there hasn't been much that made me want to reach for the keyboard, either....

Remember when I worked out that Telstra's bid for the national broadband initiative showed that they had been fully expecting to make several billions in clear profit from their shabbily prepared tender?  Reading this shows me that there's a far deeper problem involved.

It looks like every company involved in the tender thought they were going to clear billions and billions in profit...  And what can that be traced back to?  Well, I think Telstra's lazy insolent smugness at having been the monopolistic sole telco for Australia for so long, has rubbed off onto every other telco they've been involved in.  It's always been easy to stay under Telstra's prices, but even easier to keep prices high and blame Telstra for their monopoly pricing to other telcos.

Sol Trujillo leaving has been one of the best things to happen to Telstra - to date.  I won't say it was the best thing ever because now we need to watch what the new incumbent(s) do.  Letting other companies lay their own fibers has also been one of the best things to date - and one of the worst.

Because many of those telcos that laid in fiber are overseas companies, and while it seems inevitable that we'll soon have to think in terms of one global economy, that time isn't quite here yet.  (It should be, but it's not.  Long story...)  So for the moment our choice is to bleed our dollars into Telstra's pockets, or into the pockets of an Asian telco cartel, or something.  When Telstra was still a government-owned and operated department, we probably didn't mind so much because we were kind of keeping it in the family.

Now, it's unacceptable.  It's distasteful, and even worse, it's now put us somewhere in the range of third world countries when it comes to price of digital access.  I'd like to bring a quick observation to the table.  When I was running an ISP back in the mid 90's, data access was via ISDN lines, and our clients accessed our server via 33K dialup modems.  The cost of the ISDN data was actually quite reasonable.

Then Telstra realised in the late 90's that it was sitting on an untapped goldmine, and the price went up from 9c to 19c a meg and only a few years after that, the ISP sold to another ISP.  But think about that for the moment - if you have a 5Gb a month cap on your downloads, then at those prices you'd be paying $950 a month for that data.  Before allowing for inflation, which would make your download limit worth about $3500 in real terms,,,

And that 19c/meg figure is still around to this day.  Yep, it's just migrated to GPRS data rates.  So people with mobile phones with no 3G/HSDPA capability can look forward to big bills month after month, and really, data plans for mobile broadband aren't anywhere near realistic yet either.  While landline customers check their price in Gb, (and find that they are paying perhaps $8/Gb) mobile broadband users are still ekeing out their data in Mb's and finding that they are paying 3c/Mb or about $35/Gb on average, or four times as much as the landline customer.

And most of that is down to pure price gouging and "socially acceptable" price fixing between the telcos. Because that's the way it's been done "since Telstra days..."

I've noticed something else.  In eight years of having broadband, at three different places in a capital city and one in immediately surrounding areas, I've NEVER had acceptable performance from the existing broadband infrastructure.  Currently being in the "surrounding area" I have a 512K connection of old style ADSL which has a latency delay of about 5 - 25 seconds on everything.  At previous locations I've had ADSL and ADSL2 connections that barely managed 200K, or which disconnected and reconnected every hour due to line noise.  You get the idea.

The best performance I EVER got was a one month love affair with mobile broadband, where I was getting 1.5M speeds with almost no latency - but paying $30 a month for 1Gb and then exorbitant prices once I went over that...

So when K-Rudd talks in expansive terms about "national broadband networks" that will solve all our woes, I wish I could be as certain of that myself...  See, amid fears that the Internet is at the limit of what can be done with TCP-IP4 protocol, and we're approaching the bandwidth limits of what the network can sustain, I know that we will always find a way to make infrastructure do more with less, and also we will continue to pump infrastructure into the data economy.  But will our "national broadband network" be the right technology for that? They are, after all, touting a seven to eight year timespan and a LOT changes in IT in just 18 months...

So let's hope that the new "conglomerate" doesn't A) decide it's Telstra 2.0 and plans to make huge profits this round of tenders, B) decide it's Telstra 2.0 and can become a new monopoly and C) decide it's Telstra 2.0 and deliver Internet 0.1 standards...