3rd party integrations require a lot of skill, finesse and diplomacy if they are going to be successful. I recently had the opportunity to spearhead the development effort at my company with a OEM lead provider integration. Depending on the development shop you may find yourself with either a very rigid process or a more loosely defined experience. In either situation communication is the key to success.

If you find yourself in a loosely defined process like I did then you may have to deal with adjustments to your code as well as the providers. If the process is especially new you will likely uncover bugs or flaws in the process which need to be corrected. With this in mind it's far better to keep development iterative and to communicate and address issues in a timely manner.

On the other hand if you find yourself in a very rigid design process you'll most likely be provided with a firm data spec which is relatively concrete as well as a series of test cases to qualify your integration. In this case you'll need to iron out any potential issues upfront and to fully complete your design before coding. In most cases you'll want to code to the test, but also analyze your own system for any potential bottlenecks or choke points which could create and issue during testing.

In the end, regardless of which type of integration you use you'll want to include numerous methods of feedback to provide monitoring and quality of service assurance after deployment. I've found that it's often beneficial to enable debugging or verbose monitoring during the initial deployment as well as when unforeseen issues are encountered.

Inevitably a system administrator finds him or herself face to face with a worm at some point. My latest foray into the IT battlefront happened a few weeks ago with a rather nasty worm called Conficker. The worm is a sophisticated breed and proved a worthy, albeit frustrating, adversary.

Our first contact was the detection of an RPC attack on port 445 by Kaspersky's firewall solution. A system scan revealed that several machines with a Net-Worm.Win32.Kido infection as identified by Kaspersky or Conficker as the worm is more widely known on the net. The initial infection vector was an un-patched system using the vulnerability published by Microsoft in late Oct, MS08-067. The patch addressed an RPC hole which allows the worm to infect the system via File and Print sharing.

Unfortunately most client firewalls are unable to block the RPC requests and in a matter of hours every vulnerable system becomes infected. Our initial response was to conduct an offline scan via Live CD on all machines and to apply the Microsoft patch. We also disabled file and print sharing along with several unused but now activated system services.

Call it irrational exuberance, over hype or cheer leading, but don't call it a bubble. That's the take away lesson from the recent tech start up meltdown. While sites like ValleyWag delight in the misfortune of the latest gaggle of VC funded companies and spin events as the end of the world, the truth is that this has been nothing more than a simple boom / bust cycle for tech.

Nearly a decade ago as Silicon Valley hype reached stratospheric levels many early pundits used words like "historic" and "new economy". Companies that were little more than a fancy website were raising immense amounts of capital with IPOs while communications companies laid thousands of miles of fiber optic cable. It was if the Internet commerce sector could not be stopped and millions of average Joes sunk their retirement funds into companies like Flooz and Pets.com. In the end many companies went bust others suffered with inventory or bandwidth they could not sell, and many others lost most of their nest egg as the stock market fell.

I recently had a chance to review an enterprise level caching solution from the folks at Alachisoft called NCache. Our development team was in the market for a scalable solution to high database load which was sapping the life from our database server. NCache is an in-memory cache which can be distributed among a cluster of servers to both improve reliability and performance of data intensive apps.