r,
I didn't realize how much money went into support! If the big $ is in planning and design, wouldn't (couldn't) DELL get into that phase fairly easily, as that is a 'brain powered' application, and DELL certainly has the 'where-with-all' to capture the interest of some 'brains'.
As for the 'exchange' business, was DELL even competing at that time? Once a 'seat' is installed, how often are changes to the system made? Would this be considered 'old news' because there is no more spending required for the 'seat'?
>>> The 5 year cost of ownership is now about 12% hardware costs, about 35% support costs. <<<
Is the support cost at startup or spread out? If spread out, what is done under the heading of support? If startup, and I am installing a 10 system network, engineered to go to 1000 systems, does the 35% apply to just the 10 system order, or do I have to pay another 35% when I add the other 990 systems?
>>>In the old days, the equipment was not as reliable as the equipment of today
Unfortunately not true. Base level hardware reliability has risen somewhat for PC class machines (especially servers) but is still several orders of magnitude below the 'big iron'. S390 and VAX systems routinely show availability of 99.95 and many are set up for 99.99. The best managed PC based systems (using RAID, failover networks etc.) are at about 98.5. <<<
Now this is an interesting point. When I worked for IBM, there was a thing called 'scheduled' maintenance. The Customer Engineers had a saying that 'Maintenance would be performed at your (the customers) choice or the machines choice, your (the customers) option.' At which time a maintenance schedule was generally established. I spent 5 1/2 days per week in 'scheduled' maintenance interspersed with 'crash repair'. Scheduled maintenance for small systems is NOT! Heck, what would you do? (I suppose, back up the system for one thing!) Now the 'system availability' figure of 99.95-99.99 represents the time 'outside' of the maintenance allocation.
>>> The problem is not hardware, it is the ability to keep the system up during routine operations. Every time a device is added to the configuration, or a driver is upgraded, or the network topology or security is changed, the system must be taken down and restarted, and the whole system loses performance while the changes propagate. <<<
Don't the big systems have the same problem? Again, when I worked for IBM, these changes were made and tested during the 'maintenance' allocation of time. On today's system, time is no longer allocated hence when a change is made it has to be made at the cost of lost production. The new 'plug and play' concept might make these changes more transparent, but still there would be some system loss as the software reconfigures around the new device.
>>>It will be at least another 2 or 3 years before these systems can begin to approach the minimum standards that S390, VAX, and other 'big iron' achieved routinely in the mid-80's.<<<
If you counted the 'scheduled' maintenance in the overall time frame, what would the 99.99 decrease to? I usually got 4 hours per week on my System 360's for maintenance, and 4/168 is about 2.3% of the available time which would drop the 99.99 to 97.69 (99.99-2.3) which compares quite favorably to the 98.5 you mention. ( 98.5 vs. 97.69 )
In either case, they sure run better that the originals!
Regards,
Ken |
