I want to say this was all debated once in the past back in the dialup era. If you advertised 'unlimited' dialup, you had to deliver and couldn't back door in per hour charges, etc. What makes this any different?
A feature that has yet to appear in the Xeon line, and Intel claims to already have a fix to bake into the next steppings so... Opterons can go back to being scared of the future.
Don't take my word for it: http://www.cs.tufts.edu/comp/1...
This is an area where post compile optimization can shine. By watching actual execution with live data, the post compiler optimizer can build branch choice stats to tune against based on actual operation rather than static analysis at compile time. HP's dynamo project IIRC was based around this idea, it'd recompile binaries for the same architecture it ran on after observing them running a few times. I believe the claims were an average 10% improvement in perf over just compiler optimized binaries.
On the flip side, most ISP operated edge DNS servers ignore low TTLs below their desired threshold now as a means of reducing load, I've seen some force a 24hr minimum.
There are ways to do VPNs from dynamic endpoints that don't require dynamic DNS. IPSec supports xauth for just this purpose.
Come to think of it, I'm getting emails from VMWare asking for permission to get further emails from them as well...
It's not just MS, OpenSRS (Based out of Canada) has just done away with their email notification for system outages as well. They're now providing an RSS feed or you can periodically check their blog. Their solution for those who liked email alerts, a third party service that watches the RSS feed and emails on updates...
A reboot isn't a power cycle... and at least on the Intel's if you go with an enterprise model they stay in RO mode. It's certainly something to consider, I'd hope for an appliance design the estimated write volume would be taken into consideration also so you would never plan for the drive to reach that point in the appliance's life span?
The drive did go into read only, until power cycled. As documented.
I get the planned obsolescence gripe, but it didn't lock out until over twice it's advertised write capacity had been burned through, and again, at no time did it corrupt data. You light the fuse with the first write and advance towards the time bomb with each additional one, so planned or not the drive only has a finite life span. Would you prefer the Samsung's failure mode instead?
Actually, if you read the article...
None of the drives died at their 200TB rated endurance, although the Samsung DID fail a data retention test. The Intel let go at 700+ TB of writes along with two other drives, but did so with plenty of advance warning and died in a way as to allow for one last read off of the data without corrupting it with a bad write. Hard to fault them there.
If there is enough bandwidth that there is no congestion or queing required, QoS is useless.
And QoS isn't needed if you have enough bandwidth in place in the first place.
That is your opinion. And you can do what you want with your mailing list server.
And any domain owner can configure DMARC if (s)he wants to. Which leaves the recipient mailserver operator free to NOT accept the message from your mailinglist server. Your opinion is not internet-law (even if it is written in RFC).
And that is why DMARC is a bad standard that hopefully the net as a whole rejects. They purposely avoided the RFC process. RFCs may not be 'internet law' but if everyone decides to start going their own way, we're going to end up back in the olden days of IM with everyone stuck in balkanized little e-mail fiefdoms unable to contact other fiefdoms. Would sir like to sign up for Google's Internet, Microsoft's, Yahoo's? Pick one, and hope your friends pick the same.
Breaking normal mailinglist behavior... DMARC is based on a misinterpretation and misuse of email headers.