base2Services

Continuous Integration and its benefits

2011-03-17T12:57:00.003+11:00

A key attribute to developing software rapidly as possible is continuous integration. This is a practice where members of a development team integrate their work frequently; typically each developer integrates their working copy several times a day, which by extension, leads to multiple integrations per day. The integrations are continually verified by an automated testing process (which is an essential component of continuous integration) as they come in, in order to detect bugs as quickly as possible. This process of rapid integration and iteration usually leads to significantly reduced integration problems and allows a team to develop better software faster. Benefits include: reduced risk (issues are uncovered more rapidly), easier to find and remove bugs (which are usually cumulative, the sooner a bug can be found and eliminated the better), and if you are looking to deploy more frequently (who isn’t) it can help increase your overall throughput and quality in order to reach that goal.

See how base2Services can help you achieve Continuous Integration here

Apache Harmony - The Java Fork In Waiting

2010-10-25T15:52:00.004+11:00

After driving Apache Harmony by contributing the bulk of the founding code, and by 2006 12 out of the 16 developers working on the project, IBM has now abandoned Harmony in favour of the Oracle based Open JDK project. The aim of Harmony is to create a compatible, independent implementation of the Java SE 5 (6, 7...) JDK under the Apache License v2 which is comprised of a community-developed modular runtime (VM and class library) architecture.

Is is hypothesised that this effort on IBM's part, in conjunction with their support for the OSGi Alliance via the Eclipse foundation, has been to build an entirely freestanding java based enterprise platform, enabling the choice to move away from the overcooked J2EE component set called Websphere, or more probably keeping the name Websphere and changing what it means to be a set of tools centred around Eclipse Equinox, rather than the current Websphere J2EE App Server. And so the strategy goes, this platform was all going to run on the elegant, modularised, super lean, Harmony JVM. This was the line being run by IBM as recently as last month at the OSGi Community Event in London.

Then came the wrecking ball - the Harmony based Google Android Dalvik. Sooner or later the Oracle lawyers were going to get to the part of the Apache Harmony / Sun Java agreements that effectively says that Harmony cannot be put to Mobile Use via the Sun imposed 'Field Of Use Restrictions'. It seems that as soon as they discovered them the lawyers pulled the trigger on Dalvik, and that act that has brought the Java world to the brink. The historical record is clear that IBM was not happy with the Field of Use restrictions on the Harmony TCK, but given that for IBM Harmony was an enterprise play, it was not a show-stopper in going forward with the development of Harmony.

It seems IBM decided they might suddenly be drawn into a very large dispute in which they had no benefit in being forced to take sides. Hence the IBM top down push to bail out of Harmony and announce a love in with OpenJDK and Oracle. Presumably the IBM lawyers have decided this would be a big enough firewall between themselves and Oracle v Google to comfortably sit on the sidelines until it is over. But the outcome of the battle has enormous implications for Java. Either way in my view the Java technology landscape is going to permanently change.

If:

Oracle Wins :- The JVM as the the ubiquitous byte-code-platform-for-the-foreseeable-future is dead. Google will produce the Javakiller platform, based on the Register Oriented architecture of Dalvik, without the Harmony libraries. Most of the Open Source Community will go with this, including Springsource. Apache Felix will probably produce a clean implementation of the OSGi container on this platform, possibly in conjunction with Atlassian, who are tinkering with such a thing on Java today. Redhat-JBoss will probably be the Open Source standout here, and stay Java. IBM will possibly stay Java bases along with the Eclipse foundation, or they may decide to straddle both camps.

Google Wins :- Apache Harmony will fork Java at JDK 1.6, everybody will go with the fork except Oracle.

OSGi Community Event

2010-10-06T12:13:00.009+11:00

I have just come back from the OSGi Community Event in London (Sept 29, 30 2010). I have been aware of OSGi for a number of years without ever having taken a good look. base2's recent partnering with FUSE has changed all that, and I thought attending this conference would be one way to catch up on all the things that have been happening.

After the first couple of hours of the conference I realised we have got a very very great deal of catching up to do. Sometime in the next year OSGi is going to be an 'overnight sensation' - which I'm sure will bring a wry smile to the faces of the people I met at the conference who have been beavering away with the OSGi technology and concepts - for up to 10 years in some cases. So for all of you technologists who are where I was at a couple of months ago, if you want to get your head around OSGi, where it is today, and why it is coming your way soon - here is a some reading to get started with in your idle moments.

First thing you need to read are these four articles on the OSGi Alliance site. It is quite likely that you will follow some of the links in the getting started article and never get back here, but if you do there is more interesting material below.

Benefits of using OSGi

The OSGi Architecture

OSGi Technology Overview

Getting Started With OSGi

Firstly Eclipse is all about OSGi, all Eclipse plugins are OSGi bundles being deployed into Eclipse Equinox (3rd party ones at that, which is why occasionally they don't work so well - the plugin author is often creating their very first OSGi bundle). Some Eclipse projects using/providing OSGi that I know about are:

One of the two pre-eminent OSGi containers - Equinox

An OSGi Application Server - Virgo

The BIRT Reporting Application

Then there is the Apache work:

The other pre-eminent OSGi container - Felix

The FUSE ESB centrepiece - ServiceMix

An OSGi runtime to house an OSGi container - Karaf

Distributed OSGi RI (AKA FUSE Services Framework) - CFX

The JMS Provider (AKA FUSE Messaging) - Active MQ

The Enterprise Toolkit - Aries

Now - Remember Roy Fielding and all that REST - Sling

There is the whole world of stuff happening at WSO2 - dont go in there - you'll get lost -but - the Stratos Cloud platform based on the OSGi compliant Carbon looks pretty amazing; and anybody that has been looking for an Atom based RESTful Registry with comprehensive Repository capabilities, as I have for quite some time - Governance Registry ticks all my function points.

Then there is Nimble - if you have got this far you can figure it out for yourself - it looks pretty slick, I think we will be using it

Now we are getting to the extreme end of the world - actually that just means I haven't figured out what it is for yet - but Peaberry - check it out.

Good luck with all that

Neil

Answering the iPhone developer wish list

2009-03-18T15:43:00.007+11:00

The iPhone 3.0 software announcement this morning really answered the majority of wish lists in an amazingly comprehensive manner. Most of what is left; camera quality, video, video chat, battery life, is all hardware related. So we can now all look forward to the next Apple announcement in the northern summer when presumably the software will be launched alongside the next generation of iPhone.

The consumer features are getting wide coverage on a large number of sites but lets take a look from our perspective of a business considering entering the App market with consumer or business applications.

The State of the Market
Penetration

80 countries (77 now have App Store access)
17 million iPhones sold to the end of calendar 2008
30 million phones & iPod Touches total. For most apps this is the new addressable number.

In the development market there is remarkable interest with 800,000 SDK Downloads and 50,000 signed developers, 60% of them new to Apple development.

Outcomes
25,000 Apps with 800 million downloads in just eight months. The previously highly ambitious billion download forecast in the first year, now looks like a very modest goal.

From a business perspective there are now new ways to monetize development and improve returns. Strangely upgrading from a Free to a Paid version of an App is not one of them, according to Apple it would have confused the user. If you like the Lite version you need to go back to the App store to buy the full featured version. However charge $0.99 in the first place and you can then start to sell upgrade packs, subscriptions and new content directly within your app.

The In-App purchase method supports periodic memberships or subscriptions to magazines, newspapers or trade information. It allows the purchase of new levels or add on packs in an app. It can be used to unlock downloaded features or premium content or within a reader purchase new books or multimedia content.

The caution is of course the proposition still has to be compelling enough to win an audience. We now have a whole new world of discovery ahead around what the product levers are and how to maneuver them. The recent research by Pinch Media really highlights how few apps most users, use regularly. Get the value proposition wrong with these extra options and you could kill app usage very quickly as well as earn the kinds of reviews that can badly hurt your business.

In terms of the other APIs, there is a social theme to many that Apple announced today. Peer to Peer discovery allows apps to find each other locally and co-operate for multiplayer games or sharing of effort or info. You can even visualize each other’s locations with maps now able to be built into the app

The ability to interact with accessories opens up a whole range of new possibilities. For example with a barcode scanner, collaborative collection and sharing of information and pricing. With data collection instruments, logging and analyzing data in the field as well as transmitting it back to a computer.

The final release of Push Notification was welcome, as was the admission by Apple that they got the scale of what was needed wrong by a massive factor. After the MobileMe launch they were obviously very cautious and have now completely re-architected for scale and interestingly worked with all their carriers to ensure there are no bottlenecks on their networks in pushing the info to customers. I for one am happy to see them back off and fix it well before it became a disaster.

They did touch again on background apps and provided examples of what running one polling app in the background does to standby battery life. This is running the app but with no actual activity on other mobile OS’s. According to their research standby time is reduced by 80% or more. Their push notification still reduces standby time but by a more modest 23%. I still think there is a case for some background apps, for example logging geo locations on a route while using other apps. Hopefully as battery life improves they will enable some mechanism for doing some of these things.

Some of the other features you can now build into your apps are in-app email to mail content directly from you app, access to the iPod media library and use of the proximity sensor. Video streams will now also automatically scale to the connection bandwidth available to a user, so you can now provide high quality video and no longer have to focus on the lowest common denominator.

So now all you need is the next right idea

Is developing iPhone Apps a waste of time?

2009-03-04T16:25:00.000+11:00

Pinch Media recently released a presentation they made to the New York iPhone Developers Meetup on Feb 18th. Reaction in some circles would suggest iPhone App development is not worth the time and effort. Yet applying some business basics to the data shared by Pinch suggests otherwise.

I took a look at the presentation at the time and thought it offered some interesting insights, was quite logical and gave some good tips on being successful. And there is one slide (5) that gives a very interest insight into pricing to drive demand that is gold. Take a look if you haven't yet seen it.

iPhone AppStore Secrets - Pinch Media

View more presentations from pinchmedia. (tags: pinch media)

So I’ve watched other blogs cover it over the last week or so and the headlines have ranged from the sky is falling, through a couple that seem to seem to be on the same wave length as my thinking, namely there are no guarantees of success but this is a viable and growing market and you can succeed if the product, pricing and marketing are solid.

(Oh no)

App Store grows, but apps are seldom used (Macworld)

Most iPhone applications gathering dust (CNet)

Apple App Store Downloads Often Abandoned (Infoworld)

Apple App Store grows, but iPhone apps are seldom used (Computerworld)

(more realistic)

Pinch Media Data Shows The Average Shelf Life Of An iPhone App Is Less Than 30 Days (techcrunch)

Why To Charge For Your iPhone App: Ad Revenue Stinks

We know the app store now offers a vast choice of apps
We know overall usage blows every other mobile platform of the map
We know price barriers to downloading and trying apps are low and encourage trial
We know from the number of downloads that people are doing just that.

So I am wondering what I am missing that the sky is falling writers are getting?

My iPhone usage is not typical. I use it heavily. But having talked to a number of other users this last week ranging from very light users through to nearly as bad as me, there seems to be a reasonable pattern.

We all have a small number of core apps we use 2-5 times a week and rely on. How many of core apps seems to be the biggest difference between light, medium and heavy users.
We have some games that we try once or twice and discard. Some games that we use heavily for a while, then get bored with a move on and one or two we are addicted to (for the moment)
And a lot of specific situation apps and utilities that we use from once a week to once in a blue moon when the situation calls for it
and most of us could do with a clean out of stuff we tried and didn’t like.

In my case it looks like of the 103 icons on my phone, 68 are third party apps and the rest Apple installed apps or web bookmarks

Again I am not saying this is typical but it seems to be close to the Pinch data and highly reasonable. It is actually a lot like my everyday use of computer applications. One difference is the "as needed" type applications. On the phone these tend to be one off situational apps eg I need a good meal so I use urbanspoon, I want to identify a song so I use Shazam. On the computer "as needed" use for me tends to be driven by projects, so I might use an app for a week, then not touch it again for 12 months.

The very nature of apps lends them to this very specific tasks, very occasionally model. It does however reinforce the message about not expecting reward from advertising. A developer needs to be paid for the app either by the user or by a company commissioning a branded app as a service for customers.

iPhone still growing in mobile browsing share

2009-03-04T15:18:00.006+11:00

Net Applications are now measuring mobile Market Share. According to their first mobile report, the iPhone commanded two thirds of all mobile browsing in February.

Other analysts have previously credited Apple products with high US usage using various methodologies but not this dominant. Net Applications measure web use by device and browser globally for around 160 million users a month. They have become a de-facto industry measure of browser and OS share in web browsing and are undoubtedly hoping to repeat that in mobile.

The other good news for developers is that Android users seem to also be very high internet users given the installed base of the platform.

Although the iPhone has a commanding lead in mobile browsing share, Android and BlackBerry are rapidly gaining market share. This does not mean that iPhone web browsing is shrinking, because the overall market is growing rapidly. Android has garnered over 6% of mobile web browsing since its release in October.

Mobile Browsing by Platform Market Share

February, 2009

Mobile Browsing by Platform

Total Market Share

iPhone

66.44%

Java ME

9.11%

Windows Mobile

6.90%

Android

6.26%

Symbian

6.17%

Palm

2.37%

BlackBerry

2.24%

BREW

0.51%

Report generated Tuesday, March 03, 2009 3:57:39 AM

Holy App Store Batman

2009-02-11T16:34:00.006+11:00

500 in July to 10,000 in December took five months. Two months later the number of iPhone apps in the iTunes App Store has doubled to 20,000. We've touched on some the pluses and minuses of such explosive growth before, but its worth looking at some aspects again.

This is unprecedented growth, much faster than the original music store in iTunes and shows an incredible amount of global developer activity. The acceptance of the platform is amazing.

On the downside:

It presents late entrants with a much greater barrier to getting exposure. Marketing strategies gain more importance everyday.
It presents users with a real issue in working out what apps are worthwhile. Apple need to continue to improve the tools in the App Store
It puts continuing pressure on potential earnings for developers

On the upside:

Downloads appear to be keeping pace and people are actually using apps
Users are investing emotionally and financially in personalizing their iPhone creating stickiness
The critical mass of the iPhone platform continues to gather steam making it easier for businesses to justify investing in developing on a single mobile platform for sometime to come.
The success of the App Store will continue to drive iPhone sales. The size of the market becomes more viable everyday, for collaborative or aggregation type opportunities as well as just selling apps.

Conclusions

Somewhere, sometime, someone will probably say 20,000 apps is enough and there is nothing else left to build ... just before the next killer app is launched.

The market for iPhone and iPod Touch Apps is maturing. It now requires more traditional business planning, particularly around market sizing, ensuring your investment in development is appropriate for the risk and that you have a strong marketing plan to make sure you stand out.

However the world is always ready for the next great innovation. If you have got "the idea" for the next unique or highly targeted app, it doesn't matter if you are competing with 100,000 there is still room to go for it.

Here's how to become an iPhone Developer in 8 easy steps, or just talk to us.

iPhone the only mobile platform worth developing for!

2009-02-04T12:37:00.005+11:00

A couple of new reports on the overwhelming numbers and huge growth of the iPhone software would lead you to that conclusion.

Its a mobile world and things change quickly. But at this exact point in time if you are not on the iPhone and you have a general consumer application, you are not where the market is.

In mobile gaming ComScore report 17% growth last year and while 3.8% of phone users have downloaded a game, on the iPhone 32.4% have in 6 months. iPhone users accounted for 14% of all mobile downloads in 2008, bearing in mind the app store was only available for 6 months of that and grew rapidly as the year went on.

Developing for a potential market of millions of other phones is wasted when user behavior dictates that the actual market for not only games but all types of apps is mainly with iPhone users. The complexity of developing for a plethora of other platforms, screen sizes and devices and of dealing with multiple carriers adds huge costs, and for what outcome?

In a great analysis Pete Burrows from Business Week quotes Pelago CEO Jeff Holden

late last year, he crunched the numbers and came to a shocking conclusion: that the 13 million owners of iPhone owners had already downloaded as much software as—are you sitting down—1.1 billion other cell-phone owners

Growth in the last couple of months now has that at 1.6 billion other cell-phone users or 6.4 times the size of the entire U.S. mobile phone market.

That doesn’t mean the App store is perfect, pricing pressure from low quality apps is still restricting developer margins, innovation or quality. But as the market becomes more sophisticated and users more discerning, there is every reason to think that this very competitive market will actually produce a great staircase of innovation vs price options for consumers and still allow business to produce good returns. Another business week article in mid Jan looked at what it called the Apple App Store Monster and contains some big numbers.

Update 4 Feb 2:00 pm. Just came across a Fortune blog on this top with an in-depth app analysis by category. It shows there are now 18,700 Apps available.

IDC confirm 125k Aussie iPhones in first 3 months

2009-01-30T15:03:00.005+11:00

In our early posts on the size of the Australian iPhone potential we quoted forecasts from Telsyte.
Sometime in 2011 more Australians will be able to access the Internet with their phone than with a computer
Its not just the iPhone
I have to say sometimes they raised eyebrows but it tuns out they were probably in the ballpark.

ZDnet are today quoting IDC analysis that says 125,000 iphone shipped in Australia in the first 2 and a half months.

If you are thinking about developing iPhone Apps it makes sense to try and do it for a global market. However even products with a purely Australian bent do have a viable market. Maybe forecasts of a next generation smartphone installed base of 2-2.5 million by the end of the year are a little high given events in the real world but the general direction is solid and growing daily.

As we said last year we expect Android based phones and new iPhones will drive demand in the second half of this year. Having apps ready for the spotlight during those launches gives you more opportunity for success. Having locally focused products is another way of getting noticed among 15,000 other apps. If you have an idea that only works in Australia, milk it for what it is worth and don't worry about the market you don't have. Just keep your forecasts realistic.

Apple and Nokia will swap places by 2013

2009-01-14T08:55:00.012+11:00

In a week that sees the announcement of another credible smartphone in the Palm Pre and Telstra originated rumours of another Android based iPhone killer from HTC, one research company is predicting Apple will take 40% of the Smartphone Market by 2013. Nokia currently has 40% of the market.

According to a Press Release sent to iPod Observer, Generator Research have just completed a new report that estimates Apple will ship as many as 77 million phones, mainly at the expense of Nokia. Their analysis shows the key driver is that

the iPhone and App Store constitute a vertical platform for the delivery of advanced mobile services.

Back in November in It's more than KISS, I looked at the competitive strengths of the Apple ecosystem and came to the same conclusion.

As Andrew Sheehy from Generator says

Right now, Apple has the best platform and the best-looking forward roadmap

Certainly if you are ready to get into the market, and now is definitely the time to be experimenting with mobile solutions, it is the only viable platform for consumer oriented Apps.

Navigating 10,000+ Apps

2008-12-12T13:11:00.003+11:00

Apple claim “this is going to change everything”, Ed Baig in USA Today proclaims the iTunes App store the killer iPhone App but now developers are worrying about how to get noticed among thousands of Apps without dropping their price to 99 cents to get on the top app download lists.

Over 10,000 Apps, 300 million downloads, frequent updates; in short a thriving eco-system. App developers on board at the start thought they had the market sussed. But if Craig Hockenberry from the IconFactory in his open letter to Steve Jobs this week is an example there is some very fast re-adjusting of thinking to be done.

With 500 Apps, the initial version of the App Store was a hit driven machine. Top Download lists for paid and unpaid Apps drive much of the activity. In an incredibly short time we have gone from a small hit driven market and are well on the way to becoming a classic Long Tail marketplace. If the theory holds, and there is no reason to suspect otherwise, everything will sell, but the top apps will make up 50% of the sales and everyone else will have to share the remainder. With 50,000 or 100,000 Apps the marketplace itself will need to work very differently and winning in niches over time will become far more important than being the download of the week.

Business models for all but the really unique or revolutionary apps need to be different. Apple need to do some work on the market itself to allow users to sift through Apps more quickly. That is already happening in games but it needs to happen in the broad catch all categories like utilities, books and entertainment. But the marketing role is not Apple’s, it is the company selling the product.

Business plans need both a launch and a long term marketing strategy. Initial plans should include all the digital and traditional channels you would normally use to effectively reach your target market. ROI will quickly determine what channels you actually use for your available marketing dollars. The job of selling the product and price to the end user belongs to the developer not Apple. Treat iTunes as a great distribution channel and if you can tap into the extra exposure reviewers and Apple provide during your launch phase, take it as a bonus

As Peter Cohen points out this is not Field of Dreams stuff or as Anthony Franco puts it and I’m paraphrasing, get serious about creating something truly differentiating or just do a web app. Into that equation you need to factor how effective you can make your marketing.

Branding and the iPhone

2008-12-12T09:28:00.003+11:00

In addition to the fun, social and utility apps that naturally have a place on the iPhone, it has always seemed to us that there is a real place on the iPhone for the brands we deal with regularly.

That said, there has to be true value in the apps that makes an iPhone owners life easier, cheaper, more fun or more informed. True useful functionality will do more to enhance the relationship with a brand than a bucketful of straight advertising dollars. Our NearMeee platform was in fact designed to help companies get there with highly functional apps at an affordable cost.

Internationally we are now seeing a number of branded Apps appearing from large companies. Most of these are not available in the Australian iTunes store so we have to accept the word of reviewers that these are well thought out, useful Apps

Poorly designed apps and those that are just plain advertising based are likely to be a very doubtful route in this market. Unfortunately the branded Australian apps release so far are tending to fall short of the mark. If they are updated quickly no real lasting harm is done but if they sit in this state while the rest of the world move on, then brand image will suffer with iPhone owners

The Qantas App is solid but very raw. It has arrivals, departures, timetables, lounge listings and phone numbers. But it does not use your current location as a default when prompting for information and does not show flight status, seat availability or pricing, let alone allow for bookings to be made or moved. All the things you want to do most when traveling with your iPhone. 4/10

The less said about the Toyota Dealer search app the better. At least in the second version you can call or email the dealers using the data shown. The fact the app ignores your current location and makes you enter a postcode or suburb, probably has more to do with extremely consumer unfriendly business models involving PMA’s (Prime Marketing Areas) for Dealers than any technical limitation. The result though is just a laughable user experience. 2/10

State of the App Store @10,000

2008-12-04T00:09:00.002+11:00

Debate on 10,000 Apps being available on ITunes or not is really relevant to most of us, because each country has slightly different access and availability.

What is more important to most of us is the state of play for those Apps. Especially the divide between free and paid apps (22% v 78%), the prices and by category differences depending on what you are interested in developing. Ben Lorica has just done a fascinating analysis for O’Reilly Radar

Focus on Migrating your Application backbones first

2008-11-25T07:07:00.005+11:00

The current financial crisis has really started to make people think about ways to reduce there ongoing costs without spending too much capital. Some organisations are looking at moving to a Linux Desktop, others are looking at buying off the shelf packages and some are even looking at internal bespoke development to save on outsourcing.

The problem with these approaches is that a high degree of training is required of your staff and in some cases customers. Productivity is also impacted to a degree whilst people learn these new technologies. Hence, increasing your total cost of ownership.

Avoiding additional costs beyond development is key to reducing your overall spend and this can be achieved quite quickly. We have found that the best way to achieve this is to focus on the licensing costs of your applications backbone, typically the application server.

The application server requires minimal user training, it is support staff that need to understand the new monitoring tools, and no impact to the look and feel of your applications. In fact, in some cases this is as simple as moving from a licensed Application Server to an Open Source Application Server with simple property file changes.

base2Services started Migrating customers onto Open Source about 3 years ago and each instance was successful and completed in a very short amount of time. Our analysis, which usually takes a couple days, can determine very quickly the effort required. To read more about our Migration approach just follow this link.

Its not just the iPhone

2008-11-25T00:06:00.001+11:00

Anyone developing iPhone Apps shouldn’t forget the extra reach offered by the iPod Touch market. It makes a solid global base of iPhones even more attractive especially for the gaming market.

Even for apps purely for the Australian market it adds to the July forecast by Telsyte of a base of 250,000 local iPhone sales by Christmas.

How many is almost impossible to estimate accurately from published figures. We do know that iPod revenue is up substantially this year even while unit growth has slowed. Logically a substantial proportion of iPod sales are iPod Touches. Rough ballpark may look something like this 54.8 million iPod units x 4% Australian Sales x 20% sales of iPod Touches = 44,000 as of 30 September with the Christmas sales yet to come.

Functionality is slightly more limited on the Touch, location services rely totally on Wi-Fi triangulation and there is no camera or microphone, however many apps function well and offer enough power to users to be very attractive, at least in Wi-Fi range.

Obviously base2Services are taking an active interest in all next gen mobile platforms, especially Android, but for the next year the App market is the iPhone with a touch of Touch.

SOA Success

2008-11-17T07:35:00.002+11:00

SOA (Services Oriented Architecture) is probably one of the most misused terms in the IT industry. People often refer to SOA as Webservices or Integration Layers. In actual fact SOA is about having the ability to share services that an organisation has.

SOA is not an IT related problem nor an IT related solution (atleast in some cases). Almost every organisation I have seen or even worked for has decided that SOA can be reached by using a Middleware layer. And almost every one of this implementations has gone over-budget and failed or came close to failure.

SOA usually starts with the need to use information from various systems in other systems. It then evolves into a technology issue that revolves around reducing the number of interfaces into each system into one, thus in theory, reducing the maintenance costs overall. It then evolves into the need to introduce a Service Bus. It is at this time the Architects in an organisation begin there sales journey, looking for a project sponsor.

Once a project sponsor is found, the design phase kicks off, taking in some cases several months, usually by those without any previous experience and the same mistakes are made. Instead of looking at each individual service required at that very moment in the organisation, they look at services that either do not yet exist or that a need has not as yet being identified by the business. Hence the design is built in such a way that the first of the services will cost 10 - 100x times more than it should have.

The Project sponsor becomes a very unhappy person and the mood in the business starts to waiver towards the negative.

Focusing on the immediate services, ensuring the cost is kept to the absolute minimum for this set of services will prove to the Project Sponsor and the rest of the business that SOA can be achieved using technology. Try to avoid focusing on the “what if”’s in SOA, remember that services are only required if the business needs them, not if the Architect believes that they know better.

I know that this is a fairly single minded point of view on SOA and that those with plenty of vendor driven reasons for the need of SOA would argue against my post but it is based on the continuous issues I see in any organisation that has implemented SOA except the 1. This one organisation took the minimalist approach and it worked

iPhone a Game Changer for Business as well?

2008-11-15T23:57:00.000+11:00

The iPhone 3G release with the App Store and v2.0 software with business features, including exchange support, expanded the potential for iPhones in business. The collective clamor from the enterprise IT experts decided it was still not enough for Apple to be taken seriously. Gartner though gave it qualified approval in August. 5 months later what are the signs and what opportunities does it point to?

In early November J.D.Power and Associates announced Apple’s iPhone had won the 2008 Business Smartphone Customer Satisfaction Survey

In the overall perspective the trends even since the release of the original iPhone and accelerating with the 3G version is that Apple is stealing business marketshare from Windows, Palm and Symbian (Nokia/Sony-Ericsson) and becoming a serious competitor to Blackberry maker Research in Motion. Globally Nokia is still market leader in smartphones but already lagging behind RIM and Apple in the US market. According to Canalys, in the September quarter 2008, Apple beat RIM and came in #2 globally to Nokia.

As RIM is still the favorite of corporate IT in the US, Apple’s gains appear to be primarily in SME’s but there is anecdotal evidence of growth in higher executive levels of larger corporates.

If you look at the available apps in business and productivity it looks like most developers have been on the right track with apps that focus on business intelligence and personal productivity. These have appeal across these categories but certainly don’t yet fill all the needs of a mobile workforce. My guess is that these will be filled both by proprietary enterprise apps and longer term by both the open source and enterprise solution providers moving from generic mobile add ons to highly optimized iPhone Apps using the full functionality and speed of the phone.

Replacing laptops with iPhones for certain mobile workers is a real possibility when the core features are offered in a optimized form specific to a company.

It's more than KISS

2008-11-04T23:51:00.000+11:00

We all know how well Apple applies the KISS principle. What they selectively leave out of interfaces, software and products like the Macbook Air takes up copious amount of blog space worldwide. But the products survive and thrive and in time Steve Jobs is usually proven right.

There’s another principal Apple apply as well and its the one that keeps them ahead of all players. Forget a nice acronym (ITTEB) but “It’s the total experience baby”.

Hardware companies that deliver phones with a better camera and technical specs but unusable software don’t get. I wonder if developers trying to deliver fully functioning word processors on phones, or better Jukeboxes for computers really get.

Having a whole ecosystem is a massive advantage that looks like keeping Apple well ahead of the competition for a number of years. They put it in place for the iPod, improved it with video and Apple TV and adapted and extended it for the iPhone. Because it just works for any consumer in many parts of the world it delivers a total experience that is the key difference.

Its not just the App store for the phone, its the music and the video and being able to add and subtract your own collection and syncing it all together. Its the sum of the experience that is almost impossible for other companies to define and replicate in total.

Cloning the App store without getting the rest of the experience right is not the answer and that is the real issue for everyone who tries to copy what Apple will continue to innovate.

Is there a lesson for App developers? Two lessons in fact! Keep it simple and make it part of the total experience.

Shazam is a perfect example. Tap to turn it on, listen to a piece of music and it tells you what it is - SIMPLE. But it also integrates beautifully into the the whole experience. See the video of the song on YouTube, preview and buy it from the iTunes store, share it with your friends, listen to it again on your iPod, sync it back to your computer, and on a Mac back it up automatically with Time Machine.

How big is the global market for iPhone Apps?

2008-10-22T23:41:00.003+11:00

At least One Billion apps in 2009. After Steve Job’s announcement of 200 million App downloads in 102 days, TechCrunch and Roughly Drafted Magazine both calculated that in 2009 over 1 billion iPhone Apps will be downloaded through iTunes at current rates. Depending on international growth rates, RDM estimates that the 2 billionth download may even be reached before the end of 2009.

In the only public survey of users so far, 98% of owners said they had downloaded at least 2 apps in their first 90 days, 75% had downloaded 10 or more.

With 65 million registered users, iTunes is already the number one download store worldwide and the number one music retailer in the USA of any type.

By the end of 2009 the established base of iPhones users is predicted to reach 35 million people.

Proven success + high participation + strong untapped potential + 100% growth = This is a very big market

Sometime in 2011 more Australians will be able to access the Internet with their phone than with a computer*

2008-10-21T22:12:00.000+11:00

One of core the reasons the iPhone is a game changer is that it makes the web usable in a handheld form. iPhone users in Australia already consume six times more data than the average mobile user.

That doesn’t mean a standard web page is necessarily the best way to access that information. Pages that are optimized for the iPhone generally offer a much better experiences. News Ltd and nineMSN
for example have produced great optimized iPhone web experiences from their generic web data.

Best of all though are native iPhone Apps that are able to offer an even more compelling mix of performance, features and dynamic information drawn from the Internet. There are now enough of these applications to suggest that companies serious about being a major player in mobile moving forward should be seriously evaluating and developing their strategies on the correct mix of generic web pages, optimized smartphone pages and native apps before the market gets away from them.

*Telsyte Research quoted in AustralianIT on 21 Oct 2008 Page 33 with Additional Assumptions:
1) That by 2012, 90% of SmartPhones will offer a browsing experience at least equivalent to today's iPhones
2) That an owner will replace a smartphone every two years
3) Between March and August 2009 we will see multiple models of iPhones or Android phones in the local Australian market that will increase choice for the second half of 2009.

Guvnor

2008-10-03T13:43:00.003+10:00

Rules engines have, for a long time, been promising to remove business rules from our code and placing them in a format that can be easily maintained by a business savvy person who does not necessarily have development skills.

With all it's power Drools has provided a simple yet powerful language for expressing business rules. However, to the non-technical person the rules can be just plain scary in their plain text form.

So along came Guvnor!

Guvnor solves this problem by providing a browser based interface for the editing and creation of business rules. Some of Guvnors features are:

Sophisticated user and role management.
Versioning of rules
Release management
Rule execution on a provided server or in client code
Uses GWT and is browser independant.
Provides simple, intuitive rule creation and editing
Can perform both stateful and stateless rule execution
Exposes rules base via a simple RESTful interface
Tests can be create to execute scenarios
Rule analysis that can check the rule coverage
Eclipse plugin for developers

Flexing Maven's Muscles

2008-09-16T16:08:00.032+10:00

Recently I've worked with some junior guys who had been, to a large degree, left to their own devices to work on a Flex project. The guys had done a great job writing approximately 50 templates that are used to display and in some cases edit, server side data in a highly prescribed format. They had however been doing some interesting things to get around the environment that was forced upon them.

In this blog I'm going to try and describe the environment these guys were working in, the limitations and restrictions placed upon them, and the processes we put in place to help them be more productive and therefore better able to produce a quality product in the limited time they have.

The Flex template work is a small part of an extremely large project. The core of the project is to collect large amounts of advertising data and publish it to numerous formats. The templates will be used to display the data in the various formats and allow sales people to demonstrate to advertisers exactly what the advertisement will look like. The application that collects the majority of the data is being developed by a third party.

This third party provides a number of Flex libraries that must be used in the templates to communicate with the server for updating and retrieving data. In addition to these libraries the third party provides a small number of templates and server side code that allow our team to compile and execute the templates inside Flex Builder (a specialised version of Eclipse). The ability of this third party to provide our team with ALL the correct code and libraries has been a problem that I won't go into here. We have however managed to, through the processes I'll detail soon, insulate ourselves from this to some degree.

There were a number of problems in the way the team was working that we identified immediately. They included:

Poor workspace and Eclipse project management
A lack of source control
Little or no release management
No unit testing being performed
A lot of repeated code

One of the major problems with Flex is that with any project of more than a few templates can suffer from exceptionally slow compile times. Meaning that any change to any template can result in a lot of lost time waiting for templates to compile. To solve this our team did the most logical thing and separated templates into different Eclipse projects based upon the business unit the templates were designed to display. This lead to a workspace structure something like this:

Each folder represents a project in the workspace. The "supplied-libs" project is exactly what it sounds like. It contains all the libs we get from the third party. The "execution-project" contains the code that allows us to execute the templates locally and connect to the server to get real data. The "our-libs" project contains actionscript code that our team had written. This allows us to effectively debug our code and gives us reasonably quick feedback cycle.

For each set of templates (remember we split them into smaller sets so they could be compiled in a reasonable time) our young team had created a separate workspace. This meant that each developer had up to 6 different workspaces with distinct copies of the "supplied-libs", "our-libs" and "execution-project". This may not have been a problem except that none of this code was in source control.

So, given this structure the guys found it hard to keep all the workspaces up to date with the latest libraries and if they changed or added a class in the "our-libs" project they had to not only make sure all their own workspaces were updated, but also they had to literally pass the file to their team mates. Eeww!!

So the first things we did were:

Added all source code and supplied libraries to version control
Merged all the workspaces into one per developers machine

This delivered us a number of immediate benefits:

Reduced code duplication
Fewer unexpected compile errors
Ability to get a snapshot of the entire codebase at any time.

Our team was able to ensure they were always using the latest versions of the shared code they were producing. This encouraged them to refactor the code and remove duplicate code from templates into shared actionscript classes. In addition, by "closing" template projects they were not working on at a particular time they could continue to compile their templates in a timely manner.

So, now that out team can share their code more efficiently we needed to look at how we build and release our templates for testing. There was effectively no release management. Our team was using Flex Builder to compile the templates into swf files and uploading them via a web interface. This lead to a number of problems, the main one being that because we were manually managing the inclusion of libraries and shared artifacts we were accidentally compiling against out of date code.

So along came Maven! Now at first glance Maven appears to be the perfect fit for this job. It will handle our dependencies and compile our flex code into the artifacts we need.

Or will it?

For our supplied-libs and our-libs project it works well. For our template projects it is not as straight forward. Maven works on a single artifact per pom model, which is fine where each project has a single artifact. Our template projects need to produce multiple artifacts, ie one SWF file per template.

We set up some simple projects that compile and install the libraries into our Maven repository. Our template projects then depend upon thos libraries.

One of the most important things we wanted to do was ensure that developers could still use their existing tools to execute and compile templates locally. In other words using Maven must not slow down the day to day operations of the team. We created the following project structure for our templates projects.

This structure keeps all of the mxml files in a single folder (src) and allows us to create modules that Maven can compile.

The pom.xml file at the root of the project defines a parent pom and contains the majority of the build logic and the plugins used. It also defines the modules to compile and all the dependencies.

Each module has a pom which has the root pom as its parent and then defines a property called final.name which is used in the parent pom to identify the template to compile. This use of modules is very powerful and lets us produce multiple artifacts from a single project.

We now have the ability to build artifacts based on libraries (supplied and/or ours) installed in a Maven repository. Now we can set up continuous integration to build them for us automatically.

We have done this using Hudson. Using Hudson we are able to monitor our Subversion repository and automatically compile and install the libraries and templates whenever there is a change in a libraries or code. It handles the dependencies between our projects and allows us to build any or all of our artifacts. It will also tag a build for release.

We are also working on including a version and build number in our templates. This will be visible and allow users of the templates to identify exactly which version of the template they are using. This will become particularly important when testers start reporting bugs in our templates.

In addition to this we have added unit tests for our Flex code using an open source library called FlexUnit. We are coming from a point where none of our code is unit tested so we have added ateam rule which is that if any actionscript class is added or changed then tests must be written for that all of the methods and behaviour of that class. This will allow us to, over time, improve our test coverage, thus improving overall code quality.

So, by implementing these simple concepts we have made our lives easier and given ourselves a greater chance of producing quality code. The benefits we are already seeing are:

Less repeated code.
Better release management of our artifacts.
Developers have more confidence making changes.

On top of this it makes our developers lives easier and our customers happy!