Re: eBay's hot potato toss with tail between legs

Steve Dossick (sdossick@cs.columbia.edu)
Fri, 18 Jun 1999 13:25:11 -0400 (EDT)


Last I checked, Oracle has (very expensive) redundancy software, so when
one server goes kaput the other takes its place. They can do this with n
servers, in a lineup (where n-1 are backups waiting to take over if the
guy in front fails) or in combinations of active and backup servers, with
either full redundancy or load spread across db's (kinda like raid).

Yes, it's expensive. Yes, they'd need a whole extra dba or two. No
excuse for a publicly traded company not to invest the $$ protecting their
bread and butter...

Try this on for size:

GM: "Dear customer. Yesterday you placed an order with one of our
dealers (we can't remember which one). Our order processing system lost
the order because we were too stupid/cheap/lazy to buy an extra db server
and cluster them. We've lost all the orders from the past 3 days".

Guaranteed shareholder lawsuit.

-s

On Fri, 18 Jun 1999, Koen Holtman wrote:

> On Fri, 18 Jun 1999, Steve Dossick wrote:
>
> > Given that one of the features of the high-end sun hardware they have (and
> > an available option on Oracle) is true server clustering/redundancy, why
> > didn't they just buy an identical machine and have it running as a hot
> > spare? God knows they have the $$.
>
> The Online reporter article states that they have a "massive corruption of
> the data system". Sounds like a corrupt database, very much a software
> problem. And restoring a corrupt database is one of the more manpower
> intensive things one can think of. I guess that putting back last week's
> backup is not that much of an option for eBay.
>
> I've heard a database vendor explain that their usual support procedure
> for people with a corrupt database is to ask them if they *really really
> really* need the part that is corrupt. If not, they nuke the corrupt
> part. If the part is *really really really* needed, well, this is why the
> support contracts are so expensive.
>
> Throwing redundant hardware at the problem usually gives you two corrupt
> databases in stead of one when the shoe drops.
>
> Koen.
>
>