This article on infoworld
about last Summer's outages is a typical illustration of how things can go wrong in a networked business world. With the current recession, cost cuts and commoditization of both software and hardware, the end result seems to be converging towards the state predicted by the late Jim Gray
during his Turing Award speech in 1998:
"Cheap and buggy. Sometimes it will work, sometimes not, and nobody will really know why".
Given the perspective of such a world, it seems like some precautions are justified. Since it can't be repeated often enough, here is the Atomikos view on how to alleviate all this:
- Use our transaction technology to avoid data inconsistency after a failure or crash. It acts like an insurance, really: you don't need it when things are fine, but when things are turning bad you're sure glad to have one!
- Use queuing to your advantage. Avoid depending on the availability of a remote service by delaying requests until they can be performed (i.e., when the remote service is up).
- Add virtually unlimited scalability while you're at it…
The combination of technologies and techniques outlined here will increase your reliability while decreasing costs
. You save on expenses by using commoditized hardware and software. You also save on man-hours of development because our products will allow you to focus on the happy path (failure handling is automated to a large part by our software). This significantly decreases the complexity of the workflow your developers have to code, maintain and debug. Less code in turn means less bugs, so this again increases reliability. Isn't that beautiful?