On TechRepublic: 10 biggest failures in IT history
BNET Business Network:
BNET
TechRepublic
ZDNet
TalkBack 1 of 2:
Next »
These are neither new nor substantially different to existing solutions
I agree that it is definitely worthwhile explaining the differences between fault handling modes of systems but neither of the solutions you link are substantially new or different to old school clusters.
Both of these solutions are trying to make dumb applications behave intelligently from underneath. There are well understood fundamental limitations to this approach, the 'solution' looking up the stack cannot possibly predict the behaviour or failure patterns of any application, user load and data set you choose to install on it. This leaves the solution able to look after hardware failures and make educated guesses about software behaviour. This adds substantial installation and operational difficulty for a small gain in theoretical reliability. One of the solutions has an additional layer of complexity strapped around the application to try and mitigate this fundamental failing. Instead of spending company money to stick plasters over badly written applications we should be concentrating on selecting or writing applications that are written properly, are inherently fault tolerant and do not require overpriced solutions to deal with the symptoms of their architectural failings.
Posted by: Liam Newcombe   Posted on: 04/16/07 You are currently: a Guest | Members login | Terms of Use

Alert moderator to an offensive message

Subscribe to this discussion via Email or RSS

These are neither new nor substantially different to existing solutions  Liam Newcombe | 04/16/07
Corporate networks evolve  dkusnetzky@... ZDNet Moderator | 04/21/07

What do you think?

SponsoredWhite Papers, Webcasts, and Downloads

advertisement
advertisement

SmartPlanet

Click Here