Big migrations having issues because of poor planning, poor leadership, poor user acceptance?
Lots - i'd guess almost all of them. Look at all the shit people have been spewing about Vista and Office 2007. People hate change so much they don't even consider it being a good thing.
So you're of the opinion that if something doesn't meet up with claimed performance gains, obviously the purchaser is at fault and it's not possible that the manufacturer overstated claims at all?
If Intel claims that performance with new Xeons 5500 improved 35% in their benchmark X, and it doesn't then they'd have a case.
But they aren't running benchmarks - they're running their applications, in which they (obviously) aren't getting a 35% speed improvement.
Or worse, you're saying that if you actually believed a vendor, you're an idiot?
It depends. If you're talking to an Intel sales, which claims that their new 5500 Xeons will improve performance by 200% for half the energy usage for Facebook's specific use cases, and the guy as a technical executive believes this bullshit, then yes - it's mostly his fault.
You can get a two-quadcore machine from IBM for less than 10k (IBM x3650 M2).
You can get a single-core-activated Power 520 for around 20-30k. The second 4.7 Ghz Power 6 Core is disabled. Also, this machine maxes out at 16GB of memory, while the x3650 M2 maxes out at 64GB (with the reasonably priced 4GB sticks - more with 8GB sticks).
A good way for small VMware deployments i've found to be using e.G. a NetApp filer and run the VMs over NFS. Much cheaper than a real SAN, and more flexible, while still retaining the same performance.
I'm not sure if NFS is a good idea in bigger deployments, though.
Well, i can tell you that i do not want cheap, shitty hardware with no feature as servers.
This is all fine for companies like Facebook and Google that are in the primary business of running IT, and wrote software that accomodates for the shitty hardware they use.
However, other applications like standard business IT requires highly resilient, highly managable hardware which offers many features, stable parts supplies, management possibilites, and is built upon sturdy hardware that can withstand non-datacenter conditions of cooling and dust.
Actually, there is a very easy answer: just ban any node that is disrupting service for another user
Doesn't work. Just imagine a network full of hundreds of student laptops, all unpatched at Windows XP SP1. When one gets infected, everyone will get to.
Thus, the enforcement of Windows Updates and Antivirus software is very much a necessity.
Keeping a school network secure is very, very hard.
NAP solutions, such as Ciscos Clean Access Agent are a good way to ensure that basic security requirements on clients are met. Unfortuantely, if configured incorrectly it's rather easy to circumvent Cisco's stuff if configured wrong - which it is at most schools.
And then there are the "experts" that don't want to deal with NAP, circumvent it the poorly configured NAP and start spreading viruses.
Unfortunately, the only way to properly secure such a network is to use NAP in combination with 802.1x and a secure 802.1x authentication mechanism, like EAP-TLS. This can ensure security in a school network.
Of course there are privacy concerns with NAP solutions, but i don't think the complaints are valid - if you want to use your own computer in school AND the school agrees you to give you access to their network, it should very clearly be on the terms of the school. Otherwise, you can also bring your own internet connection - many laptops have integrated UMTS as an option, and almost all carriers sell UMTS cards.
Slashdot Mirror
Big migrations having issues because of poor planning, poor leadership, poor user acceptance?
Lots - i'd guess almost all of them. Look at all the shit people have been spewing about Vista and Office 2007. People hate change so much they don't even consider it being a good thing.
If Intel claims that performance with new Xeons 5500 improved 35% in their benchmark X, and it doesn't then they'd have a case.
But they aren't running benchmarks - they're running their applications, in which they (obviously) aren't getting a 35% speed improvement.
It depends. If you're talking to an Intel sales, which claims that their new 5500 Xeons will improve performance by 200% for half the energy usage for Facebook's specific use cases, and the guy as a technical executive believes this bullshit, then yes - it's mostly his fault.
Yeah, but the guy complaining here isn't a tech, he's a policy maker.
If the rich snobs fuck up, they blame Intel and AMD. Because really, why would they admit they made a mistake?
The prices are in CHF and complete configurations - not sticker prices.
IBM's POWER stuff is overpriced.
You can get a two-quadcore machine from IBM for less than 10k (IBM x3650 M2).
You can get a single-core-activated Power 520 for around 20-30k. The second 4.7 Ghz Power 6 Core is disabled. Also, this machine maxes out at 16GB of memory, while the x3650 M2 maxes out at 64GB (with the reasonably priced 4GB sticks - more with 8GB sticks).
I'm not sure what you're talking about, but our 10'000 CHF IBM x3650 M2 beat the shit out of the 50'000 CHF IBM Power 520.
I'm quite sure that an IBM DS8000 stuffed to the brim with SSD's will alleviate most of the hard drive bottleneck issues ;)
A good way for small VMware deployments i've found to be using e.G. a NetApp filer and run the VMs over NFS. Much cheaper than a real SAN, and more flexible, while still retaining the same performance.
I'm not sure if NFS is a good idea in bigger deployments, though.
You used slow SATA Disks and complain about low IOPS when doing random IO?
You'll be sure to get a job at Facebook :)
Well, i can tell you that i do not want cheap, shitty hardware with no feature as servers.
This is all fine for companies like Facebook and Google that are in the primary business of running IT, and wrote software that accomodates for the shitty hardware they use.
However, other applications like standard business IT requires highly resilient, highly managable hardware which offers many features, stable parts supplies, management possibilites, and is built upon sturdy hardware that can withstand non-datacenter conditions of cooling and dust.
They might be able to force Atari to give out their key. Which Nintendo would probably blacklist - problem solved.
Apple doesn't have a monopoly, so they can do whatever the fuck they want.
Being "free" means exactly that other people are sometimes allowed to do stuff you don't like!
Yep, and depending where you live jailbreaking is completely legal. Then, you can install and run whatever application you want to use.
Illegal in germany.
And legal in many places - not in the US though.
Most helpdesks calls are just users being obnoxious brats.
The real approach would be to deduct 10% salary for each call to the helpdesk. If it turns out to be a real problem, the money will be refunded.
This will avoid useless calls to helpdesk, and ensure that only real problems result in cases.
I prefer answering "Yes, of course! I'll send you a quote in a few days".
Makes people shut up fast.
You actually believes anyone with decision making power cares for the big picture?
So, what you want to say is that by removing IE, Microsoft has created new jobs in Europe?
That's a good thing, isn't it?
The only way to "clean" an infected operating system is to reload it from the last, clean, uninfected backup or the original install media.
Running an IDS that can monitor every single port is _expensive_.
Users? In an education facility? Good joke, you almost had me there.
Doesn't work. Just imagine a network full of hundreds of student laptops, all unpatched at Windows XP SP1. When one gets infected, everyone will get to.
Thus, the enforcement of Windows Updates and Antivirus software is very much a necessity.
Keeping a school network secure is very, very hard.
NAP solutions, such as Ciscos Clean Access Agent are a good way to ensure that basic security requirements on clients are met. Unfortuantely, if configured incorrectly it's rather easy to circumvent Cisco's stuff if configured wrong - which it is at most schools.
And then there are the "experts" that don't want to deal with NAP, circumvent it the poorly configured NAP and start spreading viruses.
Unfortunately, the only way to properly secure such a network is to use NAP in combination with 802.1x and a secure 802.1x authentication mechanism, like EAP-TLS. This can ensure security in a school network.
Of course there are privacy concerns with NAP solutions, but i don't think the complaints are valid - if you want to use your own computer in school AND the school agrees you to give you access to their network, it should very clearly be on the terms of the school. Otherwise, you can also bring your own internet connection - many laptops have integrated UMTS as an option, and almost all carriers sell UMTS cards.
That'd be stupid, it can be easily faked.
I've secured school networks with 802.1x and EAP-TLS. Works fine - and VLAN assignment works automatically, depending on the computer plugged in.