VMWare released VMWare Fusion 3.0 on October 27, 2009. Among many enhancements and features, the most anticipated is compatibility with Max OSX Snow Leoparf 10.6 - specifically VMWare guests become 64-bit processes in this release.

After they resolved a few minor problems with their site, I went straight into buying the upgrade since I didn't have any doubts about the stability of the upgrade. After downloading and a brief setup, I started my existing VMs which immediately ran much faster.

My MacBook Pro's CPU use is a little lower and the system is cooler than it was before the upgrade while the VM is running.

 I'm really happy with the improvements in Unity mode and overall speed - great job VMWare!

When calling a receive pipeline in an orchestration's Expression shape, the receive pipeline must be a Disassembler and therefore implement the IDisassemblerComponent interface. 

Your custom pipeline won't work if isn't a disassembler - period. Although the means of calling a pipeline from within an orchestration suggests that the pipeline must be a Disassembler, since you need to call the pipeline's GetNext method, it isn't clearly stated anywhere that I looked.

I suppose it's a logical design; however, the approach forces developers to take an all or nothing approach in that you must either write your own disassembler or contain the existing disassembler within your pipeline. It would be useful to be able to call other types of pipelines from inside an orchestration.

Using Parallels 4.0, a desktop virtualization product, has been productive. The interface is great, performance has never been a problem - my Mac easily handles two VMs running at the same time under Parallels 4.0.

 An intermittent networking problem surfaced recently. Guest operating systems like Windows 2003 Server and Windows XP had trouble connecting to local resources when connected via Parallels Bridged networking option. Windows would sometimes report a network resource (file share, TFS connection, intranet site) inaccessible, non-existent, or unavailable. I could connect after several tries, but the connection wasn't reliable and would often drop during various operations.

Owning VMWare Fusion too, I converted the Parallels VMs that were having problems into VMWare Fusion (took about 30 minutes per VM). On booting up the VMWare Fusion VM in the exact state I left it in under Parallels, I immediately and successfully connected to all network resources. I continue to work trouble-free.

 I tried to post this on the Parallels support forum, but it is moderated and my post has not been approved yet.

I like Parallels, especially since it directly supports Mac Snow Leopard 10.6 for the host operating system (VMWare provides experimental host support for 10.6). I'll switch back once they fix this problem.

You may know that I do all my work using MacBook Pro. Like other notebooks, the MacBook goes into sleep mode when you close the lid/display or tap the power button and choose the Sleep option. Apple likes to make things seamless for its users - this is the main reason I switched from Windows as my day-to-day operating system (I use Windows for all of my other work, but the split between the time in Windows versus using MacOS directly is actually about 50/50!).

Apple setup Sleep such that the computer can run out of battery power when it is in Sleep mode and still be able to recover. It does this by writing the contents of RAM into a file before it goes into Sleep mode. The MacBook Pro is a fast computer, so it takes about 10 seconds to write out the contents of RAM to a file.

While this is a nice feature, the 4Gb of disk space this uses comes in handy, especially since I have a few virtual machine images ranging between 5 and 60 Gb - I can use all the disk space I have.

SmartSleep is an application that allows you to set the Sleep mode such that the Mac does not write the contents of RAM to a file before going to sleep - saving time and disk space. SmartSleep also provides a Sleep mode called SmartSleep that lets you set a number (interpreted as percent of charge left or run time on battery remaining) - once the batter hits this value, SmartSleep writes the contents of RAM out to a file to avoid loss of data! Cool.

Granted, you have to be careful about using SmartSleep in SmartSleep mode since it's possible that you MacBook can run out of battery power when it's in a nicely padded laptop bag, or being bounced around in the trunk of your car - but this is a small consideration when you think of the amount of time and disk space you'll save each time you put your MacBook to sleep (in my case, I restart my MacBook every few months, so this is really handy).

wWorkflow.net and the underlying consulting services company is please to announce that we are now buying carbon offsets to offset the carbon that gets generated as a result of developing solutions for our clients and running this site.

 A “carbon offset” is an emission reduction credit from another organization’s project that results in less carbon dioxide or other greenhouse gases in the atmosphere than would otherwise occur.

For example, assume that one notebook computer uses 75 Watts of electricity per hour for two weeks (at eight hours per day). The total energy use for the two weeks works out to about 6,000 KiloWatt hours (that's 75W times 8 hours times 10 working days). Generating about 5,500KWh emits about one tonne of Carbon into the atmosphere.

We buy carbon offsets by the tonne to offset the carbon that gets emitted into the atmosphere. By buying these offsets, we help to develop cleaner energy sources and help produce electricity using renewable sources (like solar, wind, hydro, etc) that offset our emissions. The tonne that gets emitted when the conventional electricity gets generated is already in the atmosphere, yet carbon offsets can also work to remove carbon by investing in farms (a farm removes carbon from the atmosphere when the plants it grows absorb carbon as they grow).

We buy carbon offsets from well-known, Canadian offset retailers that meet stringent standards for audit and energy generation. More information is available from the David Sizuki Foundation.

Our clients don't have to do anything to gain the benefits of carbon offsets - wWorkflow.net buys carbon offsets* each month at no cost to our clients!

The end result is a better environment, today and in the future. wWorkflow.net - providing expert and new greener solutions!

 

* So, how much does it cost to offset one tonne of carbon emissions? We operate in Ontario, Canada so one tonne of carbon offsets costs about $38 Canadian. The average Canadian resident generates about 22 tonnes of carbon per year, putting the cost to offset that at about $815 Canadian. The money gets used by reputable offset retailers to buy renewable energy that gets distributed on the regular electricity network.

MakeMyTweet is new, free service by my consulting services company.

MakeMyTweet offers free systems integration architecture and design services - no strings attached!

Visit MakeMyTweet to try it - it's free, open, and...well...free!

I'm happy to announce that my consulting services company is sponsoring part of the Stackoverflow Dev Days Toronto Conference (link).

ArtOfBabel.com is the department that's sponsoring the event's coffee breaks, so I'll be at the venue in the exhibitor's area and at the coffee breaks. 

I am giving away two tickets to Stackoverflow Dev Days Toronto - details here.

 

Most developers deploy their BizTalk 2006+ applications into other environments like Test and Production using and MSI file that they generate using BizTalk Administrator.

The process of creating an MSI is fast and easy - it's also very thorough in that the process enumerates dependencies while it creates the MSI and ensures that the dependencies are present when the MSI runs. 

The problem is that the MSI you generate on your development system could include dependencies on your system in folders and drives that may not be present on the target system. Common dependencies are maps and schemas - it's possible to end up with errors when trying to run an MSI that folders and drives don't exist!

Fortunately, there is a relatively straight-forward fix:

1. In BizTalk Administrator, under your Application's Resources folder, review the values under the Source and Destination Location columns - you'll see folders and drive letters that exist on your system.
2. For each item in the list, right-click and select Modify
3. Change the "Destination location" - the path part of the location to %BTAD_InstallDir% (include the % symbols and separate the last % from the file name (probably a DLL) with a back-slash character ( \ ). So, it might look similar to this when you are finished: %BTAD_InstallDir%\MyStuff.DLL
4.  Export your application
5. Run the MSI on your target systems

 It turns out that %BTAD_InstallDir% is the name of the folder that the user selects when she runs your MSI file - oddly enough, the installer does not use the folder for files in the Resources folder of the BizTalk application! Although %BTAD_InstallDir% looks like an environment variable, you won't find it. The pseudo-environment variable exists only within the MSI's runtime environment, so there's not much you can do to manipulate it during development.

I hope this saves you some time.

I have been working with Amazon EC2 to run several instances of Windows 2003 Server to test and develop a solution. I am using Windows 2003 Server on Amazon EC2 instead of running several virtual machines locally because:

• I want a usable system - my computers have 4Gb max RAM, so I can run two instances per computer but things get really slow
• I need to understand networking issues - local VMs are all on the same network, so it's easy to get things running (this is usually good, but I need something that's more realistic)
• I am testing scalability - Amazon EC2 can start new instances as necessary (and stop them too), based on overall load - I can't simulate that with local VMs
• I don't need durable VMs - my VMs are generic enough so that I can just start one and use it - not much configuration is necessary. It really does not justify the 40Gb of space I need for each instance

So far, I have been configuring instances to refer to each other by name (the NetBios name) by editing the hosts file. I wanted something more robust, so WINS is the most logical choice (combined with DHCP).

After much experimenting with the default, small, non-authenticated instance, I was unable to get WINS or DHCP running. While you can use DHCP and WINS on EC2, you need to use an instance of Windows with Authentication Services.

While Windows with Authentication Services implies that authentication refers only to users, it actually refers to all inter-instance communications that require authentication at the system level. 

So, why this post? It might save you a few hours of testing time since the EC2 documentation is not clear about this point. 

 

 

 

What is a Wordle? Here it is:

Click the picture to generate your own!

Thank you to everyone that expressed interest in my last article at ArtOfBabel.com - nServiceBus: Building the Solution

The article got the most in a single day so far! 

Your feedback is very helpful and I appreciate your constructive comments. 

I just got nServiceBus working on Amazon EC2!

There are two instances:

• Instance 1: Hosts Client and Distributor
• Instance 2: Hosts Servers

 

I am using the Distributor and Full Duplex samples with very few configuration changes. 

Why is this good? I have this working between two Amazon EC2 Windows instances on the Internet - so I have MSMQ working between two distributed systems that are not part of the same network.

Article to follow soon!

 I just published a new article - nServiceBus: Building the Solution

 

The article walks you through getting the latest version of the code and building it on your system.

The article also includes the latest information about the Continuous Integration server that was mentioned on this group some time ago.

 

Get the full article here: nServiceBus: Building the Solution

 

Art Of Babel is seeking articles about integration technologies, techniques, and approaches including EAI, BPM, SOA, ESB, and a broad range of other three letter acronyms.

Interested? Write a 200-250 word synopsis of your article and send it to articles@artofbabel.com This e-mail address is being protected from spambots. You need JavaScript enabled to view it . We’ll acknowledge all synopsis and may even pay for your article!

Details and other information here.

 Update:  Response has been great! I can hardly keep up - thank you!

It's official - ArtOfBabel.com is open!

Here are some pointers to the more popular articles:

• Open Source Integration with nServiceBus
• Top 4 IT Career Tips to Cope With Today’s Economy
• Understanding Publish Subscribe in BizTalk (revised)
  
• Call For Articles and Guidelines

Consider writing for ArtOfBabel.com - there's no risk (send your 200-word synposis to articles@ArtOfBabel.com) and I might even pay for your article!