*facepalm*

No, you don't use main memory as a write cache, that would be idiotic. You use it for read cache. Any updates will be written to disk (and memory) before acknowledging to the client.

In any case, it doesn't matter how good your batteries/UPS are if your server or DB software crashes for any reason and you have uncommitted writes in memory, hence why caching writes on a database is a Bad Idea (TM).

"Veteran network admin trait No. 2: If we don't know it's down, it's probably not down"

Bullshit. Well, either that or our network admins are numpties. We had a switch go down in a data hall. We knew the switch was fubar as we'd lost connectivity to a number of devices at the same time. Networks wouldn't admit a fault. We eventually got moved to new switch ports and lo, everything started working.

The usual approach to any network fault in our place is "replace your NIC". That has, in my memory, fixed maybe two faults.

The issue about ignoring future proofing is that you can invest a lot of time & effort integrating the tool into your environment, writing scripts etc. If that tool gets obsoleted for any reason, it can be a lot of work to switch to an alternative (this goes for FOSS & commercial software equally). You can get locked in to FOSS just as easily as with commercial, you just have a few more options available with FOSS. Some tools can be swapped in & out at a moment's notice, but if you integrate something into your way of working very closely, it can be a nightmare to unpick later.

As for modifying code yourself, that requires a whole set of skills & disciplines many admin teams don't have - I know our team/organisation would struggle with that. There are, of course, 100s of companies who are quite happy downloading source & patching/maintaining it themselves.

Other than that, there's some good advice in there. Version control & release/test cycles are key for any software product.

V1280s are sun4u and don't support LDOMs.

I've done Unix admin for years. I still find it quicker and easier to manage clusters with a GUI because frankly, they're too bloody complicated to manage effectively without one. Yes, you can automate with the CLI (I've written scripts to automate service group creation in VCS), but for a quick dive in to check the cluster status & configuration, it's usually quicker with a GUI to drill down to the setting you need.

Facebook & Google have networks/systems designed to work around failure and data loss is a minor inconvenience. They expect to lose a data centre at various times and continue to Just Work. In those environments, cheap grey boxes are fine provided you design appropriately. If you are designing a critical 24x7 system which cannot spread around in the same way (e.g. financial institutions) may have different requirements.

Now, while I'm not saying that Alcatel is less reliable than Cisco, Cisco generally has the reputation of reliability (warranted or otherwise) and so commands the premium.

And they didn't tend to sue people about it as much as Oracle does.

I have, albeit with Veritas. Took about a month of restriping 20TB to change the stripe widths. Just because you don't do it, doesn't mean that others don't.

Can't be perl. It's far too readable, for a start.

Site was slow for me (not quite slashdotted), but got through OK.

See Wikipedia on Submarine patents. Notably:

The ruling was upheld on September 9, 2005 by a three judge panel of the U.S. Court of Appeals for the Federal Circuit under the doctrine of laches, citing "unreasonably long delays in prosecution"

So, it seems there's a chance that waiting too long can invalidate your claims.

Compare with trademark law where you have to defend it whenever it may be seen to be infringed (see the case where Hoover corp lost the right to have the exclusive rights to the term "hoover"); the same doctrine should apply for patents. Of course, the whole patent system is a mess these days as it was designed in a different age with different industries. Scrapping patents isn't the solution as they provide valuable protection to inventors who put effort into designing something, but they're horribly abused by various parties.

The issue seems to be fundamentally down to someone with DBA rights on a database issuing "ALTER DATABASE BEGIN BACKUP" which then causes an "interlinked" database to also increment its SCN; anyone know what the "interlinking" is? I'm guessing DB links but it's a bit vague on details and high on the scaremongering... FWIW, the ALTER DATABASE command will require DBA rights to implement, so I'm not seeing the apocalypse that Infoworld is punting; if you've got DBA rights, you can do lots of stuff like drop user, drop table etc, etc, etc...

At work, we're being told our website is "in the cloud". To which I respond "you mean it's on some external webhosting? Like people have been doing for over 10 years?"

"Cloud" is being attached to all sorts of things to make them sound hip & trendy.

Wrong - if it was in userspace, it would be tied to the permissions granted the logged on user. I'm not 100% sure, but even as admin, UAC should still have blocked the worst of the behaviour. Once you're running code in the kernel, you can pretty much do whatever you want and the user's permissions and UAC become irrelevant.

If they've released the code under GPL and you still have a copy of it, you're entitled to do what you want with it, up to and including rebranding it and maintaining it as a GPL product for the future. The GPL granted you permissions to do certain things (copy, change, distribute the code) under certain conditions (you had to provide source code if requested). As far as I'm aware, they can't revoke those rights unless you break the other conditions; see Open Office/Libre Office for a similar situation.