On Sunday 26 June 2005 15:18, Trevor Warren wrote:
--- sherlock@vsnl.com wrote:
In fact i have stopped talking price with customers (i collect the M$ tax u see) and focus on a zillion other things which shreds everything that billy baba and 2000xp chors can throw.
[snip]
No worries JTD, as long as we have IIM babus running commercial organisations like those at redmond such reports are bound to surface.
Infact, the report did give me a lot to ponder. You know terence, managing networks with 500-1000 nodes with MS can be a real nightmare especially when patching the machines on a daily basis is more of a norm. MS has an architecture in place seamless across distributions to push these patches across the entire enterprise. How much of this works....you know better.
they pretend to have such an architecture in place. On friday a bank had to inaugurate two branches down south. Internet ex plodder would not connect to the central server, although routes and ports were absolutely fine. so check the router, check the nics, call up regional hub at 0230 hrs. reinstall win2k. Finally re-reinstall win2k on about 25 machines with a cd from another branch everything works. The cd that came from the software repository has the latest patches and was supposed to be installed on all machines but was delayed for some reason. U can imagine the catastrophe avoided by shear luck.
With FLOSS in a network as big as the one consisting of 1000+ nodes you will realise that management-patch scheduling-upgrades are the most significant part of the TCO.
snip
Having worked with numerous distros i still feel the need for ubiquitous patch management mechanisms is a must. Redhat with RHN comes close and but again isnt seamless across distros, dpkg and yum are other options which need to still scale to enterprise requirements. Thus until this problem continues to prevail we shall have organisations moving away from a total migration to FLOSS.
But this is a IT managment issue. Standardize the hardware first. Next standardize the distro. And have tight SLAs. Without these even FLOSS cannot generate a golden bullet. But IMO a regional repository is extremely useful in driving down costs. The bottom line is that M$ will eventually come down to a capital cost on par with FLOSS so start focussing now on the things that they simply cannot compete against.
rgds jtd