If you travel on long haul flights more than a few times, you get used to the routine. The plane announces its descent into San Francisco, Houston, London, Paris or wherever, about 30 minutes before you land. Everyone on the plane busies themselves with filling in forms and preparing for the ordeal of the passport queue. When the plane actually lands, just about everyone is desperate to get off the plane.

I went through the the same routine this morning on the way in to London Heathrow. The plane stopped at the jet way and everyone leapt from their seats and pulled their carry-ons from the overhead bins. The aisles were clogged with weary travelers in no time.

The Curious Euphemism

What could be more surprising at that point than to hear the stewardess announce:

“Could all passengers sit back in their seats please, and remove your belongings from the aisle. A welcoming committee is coming on board for one of the passengers.”

How surprising. And how graceful of all the claustrophobic passengers to sit meekly back down in their seats. What could be happening?

Perhaps Elton John was traveling incognito and some important fans wished to welcome him back to Albion’s isle. No. Maybe it was one of those dashing young princes traveling incognito and his British body guards were, well, changing guards, so to speak. Or perhaps it was some US Senator that the British Foreign Office was eager to rush through immigration so that he could have breakfast with the Queen. What could be happening?

None of those things, as it happened.

Two of the most imposing British policemen I’ve ever seen, packing heat and in no mood to negotiate, strode onto the plane and collared some young back-packer. I somehow got the impression that this was not the kind of welcoming committee the back-packer had been hoping to encounter. Once they had exited the plane it was suddenly as if nothing had happened. Everyone back on their feet and pressing forward to the exit.

I saw the young back-packer once more as I walked from the jet way. He was sitting down at a table, surrounded by his triumphant welcoming committtee – about 8 people; officious looking officials and a handful of representatives of Her Majesty’s Constabulary, including the two who had made such an unexpected entry onto the airplane.

Welcome home!

Every now and then you will see some statistic that tells you that all is not well in the configuration management processes of the typical corporate data center. It will suggest something like this:

Survey reports that 60% of downtime in the enterprise is caused by configuration errors.

In general such surveys are reasonably reliable. The simple truth is that configuration errors are common. So what causes the configuration errors?

A Repeating Theme – in 7 Steps

There seems to be a repeating theme in IT. It works like this:

1. Some new compelling development software (languages, platforms, etc.) emerges.
2. Corporate IT looks upon it and decides that it is good, and hence adopts it.
3. Applications get built using it and they are deployed.
4. Because there is no deployment framework for the new environment, deployment is manual with scripts. Errors are made.
5. To compensate the operations staff write standardized scripts to deploy and implement upgrades for the new applications. They roll their own solution
6. Eventually a script management problem emerges, because so many applications are being deployed and upgraded in this way, and the process is semi manual. The roll-your-own solution is often overwhelmed. The result is configuration errors leading to down time.
7. Eventually some software vendor emerges who has built configuration/operational management software for the environment and their product starts to sell.

This is what has happened recently with the JavaEE applications that are built to run with the major web servers. The configuration of these Java apps and applets has started to spin out of control. Consequently, a new vendor, Phurnace Software, has appeared with automation software to address the problem.

Phurnace Deliver
Phurnace’s software, Phurnace Deliver, provides a consistent technical process for deploying JavaEE applications in a way that minimizes the possibility of error and provides for swift error correction when errors are made. The way it does this is sensible, logical and easy to understand. The diagram illustrates, in outline, the way that the software works.

First of all, Phurnace Deliver stores server profiles (including the profiels of virtual servers) and also stores a complete record of all configuration scripts used to deploy, and later, update web applications written in JavaEE and deployed on the various popular web servers. For each application, users start with a baseline configuration making incremental changes. Separate scripts will, most likely, be stored for testing and test environments as distinct from operational environments.

Phurnace Deliver keeps an audit trail of all changes made so that a configuration can be rolled forward with complete knowledge of what has been changed, then easily rolled back if that proves necessary.

The software give the user the capability to contextually examine an environment and report changes that adding a new payload will make in that environment. Users are able to see the full picture. They can contextually apply a payload through environmentally sensitive updates. It is possible to freeze elements of any script to prevent them from being changed. When errors are made, users can roll back changes made and, with the help of logged information and a complete audit trail of changes, determine the root cause of the error.

In sites where Phurnace Deliver is deployed, the rate of configuration errors in JavaEE web applications is considerably reduced, as are the associated costs. It’s what you’d expect from automation of this kind. Error rates are cut and productivity is improved.

Phurnace Deliver currently has no real competition in the space it is beginning to dominate, although doubtless competition will emerge

Then at some point another development environment will emerge and become dominant and doubtless the whole cycle will repeat again, with another configuration management product emerging to fill the need.

Apple is an unusually difficult company to read. Most other companies telegraph their punches, but Apple does the opposite. It even misleads Apple watchers, so that when it releases a product it will be a surprise, in some way at least. This strategy has served it well in recent years. Before any major Apple event, like for instance the World Wide Developer Conference (WWDC) that just took place, there is a rolling wave of prepublicity that builds up like a tsunami approaching the shore line. The consequence is that the event and the announcements at the event are  well reported, even when there is very little to report.

The Message of the WWDC

The WWDC actually contained three important announcements:

1. The inevitable new iPhone (or super-upgrade) appeared, the 3GS, with much improved camera and with video capability.
2. Snow Leopard will be released in the fall and will be very cheap, at $29 for an upgrade to Leopard.
3. A 13-inch MacBook Pro was introduced and the price of  other MacBook Pros and the MacBook Air were cut.

That was it really, except for a few minor announcements for the business market. Apple will support Microsoft Exchange – it’s about time – but it also means Apple now takes the business market seriously. It didn’t use to. Another sign of this was a full 64-bit version of Snow Leopard at $499 for the server, with unlimited client licensing.

The iPhone

Apple doesn’t need to run too hard with the iPhone to stay ahead of the competition, but it’s keeping up a brisk pace anyway. The latest challenge from the Palm Pre may be real, but it’s not threatening. Apple has the App store and (see How and Why the iPhone Changes The Game), the App Store is the trump card. Developers don’t switch easily from one platform to another and Apple has the vast majority of developers for phone apps. The competition has an awful lot of running to do to get anywhere close to Apple.

Nevertheless Apple took a small swipe at Palm by cutting the cost of the cheapest iPhone to$99 and it also added some nice features like:

• Voice control
• Some neat camera features including auto-focus
• Editing video and direct publishing to Youtube
• Electronic compass – combining with map software
• Encryption
• Support for 7.2 Mbps HSDPA 3G where available

In the US Apple’s iPhone is still constrained by its exclusive deal with AT&T which will surely be dropped when it runs out. It’s doubtful whether Apple will want to make such deals again. It was good for Apple at the time of the iPhone launch and it proved to be a sensible move because the high price of the AT&T phone deal didn’t deter too many buyers. But such deals are now acting as barriers to sales of the iPhone.

The Snow Leopard

Snow Leopard will be released in the fall and will be very cheap, at $29 for an upgrade to Leopard. Why? Nobody seems to be asking that question, but I doubt if it’s prompted by trying to compete with Windows 7. Apple zoomed past Windows long ago with the move to Intel and Microsoft is clueless about how to compete. But the fact is that Apple is willingly forgoing hundreds of millions of dollars in revenue. It could easily have sold Snow Leopard at $139 and no-one would have complained.

The logical conclusion to draw is that Apple is eager to pull its customer base onto Snow Leopard, which in turn implies that it has something else in mind. My guess is that it all has to do with the coming Mac tablet. We need to consider the whole Apple range in order to get a perspective on this.

The Mac market
The third important announcement at the WDC was the repositioning of the MacBook range, introducing a 13″ MacBook Pro, cutting the cost of the other MacBook Pros and the MacBook Air and reducing the MacBook itself to the one basic model. My expectation is for that model to vanish when Apple’s new tablet computer is announced.

The reason that Apple is so tough to compete with at the moment (even in this economy) is that it has all its ducks lined up. From the iPod Shuffle up to the largest Apple server we have, in effect, a single OS, a single provider of hardware and a standard approach. If you think of this as a range of devices then the range is:

• iPods
• iPod Touch
• iPhones (iPod Touches with 3G phone capability)
• MacBooks
• Macbook Pros
• iMacs (including the Mac Mini)
• Mac Pros

The indications from the WWDC is that the price range between $500 and $1000, i.e. between iPhones and MacBooks is about to be filled by a Mac tablet. If you want to know the probable spec of such a device, read 10 Pointers To What Apple’s Netbook Will Be Like. But the device  itself will, in my opinion, only be part of the story. Apple is currently preparing to build a huge data center in North Carolina with the intention of investing hundreds of $millions – and it can’t just be for serving iTunes.

Most likely Apple intends to build on the success of iTunes and the App store and become a hub for everything digital, including ebooks and apps. Most likely Apple will expand the App store to serve apps to its soon to arrive tablet and also to Macs of every variety. Why would it not do that?

In order to pursue this with maximum impact, it probably needs as much of its customer base as possible on Snow Leopard. That, I believe, is why Snow Leopard will be cheap – to help Apple build out its full range of devices and help it to gather revenues elsewhere. Apple has a large part of the software developer community eating out of its hand. It may was well unleash them across the whole range of Apple devices.

In the end, it’s all about the ecosystem.

About a year ago I wrote Ten Reasons Why Twitter Rocks It’s time to provide a counterbalance. Nevertheless, I should be honest here and admit to being a confirmed tweeter. So if anyone working for Twitter reads this, please take it as constructive criticism. Otherwise consider it to be a vintage whine. Here are 10 aspects of Twitter suckiness:

1. It’s a jungle: I have no doubt whatsoever that there is a vast host of valuable information being injected into twitter second by second. The problem is that there is no way for you or I to mine that information in a useful way. It’s like a huge jungle filled with creatures of every size and variety making every kind of noise they can make, from screams of anguish to grunts of satisfaction. All the points below are variants on this theme, so I guess you could say this is the summary. But it’s really frustrating to me. We need to cut down the jungle and replace it with well farmed land. (I guess that’s not a politically wise allergy to use these days).

2. You can only broadcast: It’s true that on Twitter you can engage one other individual in private conversation, but only if you are following them and they allow you private contact. There’s no possibility of conferencing between 3 people. It means that you have to change channel if the need for conferencing arises.

3. The broadcast is ineffective: When you broadcast you have no idea who is going to read the tweet. Let’s say you’ve just had a sex change and want to announce it on Twitter. It’s the kind of thing that Twitter ought to be useful for. You really don’t want to surprise people you know with a gender switch, so a bit of forewarning goes a long way. Let’s just imagine you’ve “taken a trip to Trinidad” and had your warhead dismantled. You tweet “John is now gone, so call me Louise please” or something of that ilk and you expect everyone to get the message. But they’ll only get the message if they happen to be reading twitter at the time. It’s quite possible that no-one will get the message because they’ve all got their tweeting eyes on a hurricane in Florida.

4. You have no idea who reads your tweets: And neither in fact does Twitter. It just sucks those tweets in and sprays them out. You cannot mark the tweet in any way that will indicate that you want it to sit in someone’s tweet stream until it is, at least, displayed. Neither can you stop it being shown to someone who follows you, unless you go private. You have no control.

5. You have no idea what impact the tweet had. If you have a name like John Smith, it’s hopeless. Don’t even think of announcing your sex change. Even if you have a name like mine  (Bloor) which is uncommon, there is no way you can track down all the tweets that mention you, because Bloor is also a district and street in Toronto, even discounting all other Bloors. So if the word’s out that “Smithy had a sex change” and it gets retweeted many times, it’s just going to create confusion. Which Smithy, which direction?

6. Twitter has no useful usage stats. I’d love to know when the best time to tweet is, but I have no idea. Let’s say I want to get the maximum coverage for something I say. I’m a marketing guy and we’re launching a new gender enhancing product. I want to know when to tweet the fact and also how many times to do it (without annoying people). Nobody has any idea about such things. It’s a crap shoot. Tweet and be damned.

7. Following is ineffective: When you follow up to about a hundred people you get a digestible tweet stream (depending on how frequently they all tweet) but beyond that it’s just a galloping thought stream. I no longer even read the full tweet stream – it’s like trying to get a drink of water by lying on your back on a raft under Niagara Falls. It’s impossible. But to follow more than one group I need to set up multiple identities and then I have to manage multiple identities. That’s hopeless. I’ve tried it. Don’t bother.

8. The hash word is hopeless: The recent enthusiastic use of the #topic (such as #3wordsaftersex or #3wordstobreakup, etc) is useless because there is no formal taxonomy that anyone can refer to. It also consumes characters in you tweet – and it’s worth making this point en passant – so does retweeting. Why not a retweet button and make the @part separate or is that too innovative for Twitter. Because of the character limitation, people have taken to dropping the # itself which defeats the purpose and increases the noise to signal ratio.

9. Long tweets need handling better. At least we got past the constraint of 140 characters (with Twitlonger) which can be a real drag because sometimes just a little more is required. And while I’m at it I may as well acknowledge the very rare appearance of the Fail Whale nowadays, which means that at least Twitter has learned something about writing software efficiently or where fry’s is so it can buy another server every now and then. Twitlonger is better than nothing, but not much. Why not just embed the Twitlonger capabiltiy.

10 Twitternesia: After about 20 days Twitter deletes every tweet you’ve made unless you marked it as memorable. This is so sad. There are many interesting people on twitter saying, on occasion, quite memorable things. It is all being gradually lost unless there’s someone deliberately collecting it. And it’s not as if it would be so hard for Twitter to provide a logging capability. Does disk space really cost that much anyway? Google give you gigabytes of it for nothing

Having written all of that, I think Twitter is worthwhile and I’ll continue to use it. I’m just not sure why.