Forgot your password?
typodupeerror

Comment: Airplanes/cars/whatever (Score 1) 586

by Enry (#47415075) Attached to: Normal Humans Effectively Excluded From Developing Software

Remember when you could build your own airplane, or build your own car, or maybe your own radio set? Well I don't, but you could. Heck, people built their own computers for the longest time (some still do).

But the nature of just about everything is it gets more and more complicated until it's much easier to just get something prebuilt than it is to do it yourself and those who choose to do it themselves are doing it either as a hobby or because of their employer.

I've been writing code for 20 years though I've primarily been a sys admin. There are things that are much more difficult but many of the tools I used in the early 90s (bash for example, or C) are still around and follow much of the same rules as now.

Comment: Already own one (Score 1) 427

by Enry (#47320629) Attached to: Ask Slashdot: What Would It Take For You To Buy a Smartwatch?

I keep my phone in my pocket and usually on complete mute. So when there's a phone call or meeting reminder I don't get it until it's too late. Broke down and bought a Pebble a few months ago and a slight buzz at my wrist tells me there's something I need to pay attention to, and in a meeting or with friends it's a lot easier to just glance at my wrist to read a text than pull my phone out, turn on the screen, enter my unlock code, get into the app, and read the message.

Comment: Re:I just want to know (Score 5, Interesting) 538

by Enry (#47290237) Attached to: Teaching College Is No Longer a Middle Class Job

I've worked in IT at two major East Coast Universities for the past 12 years. There is a boatload of bureaucracy to be sure at almost all levels. Then again, some of it is warranted. Gone are the days of a researcher just getting a grant and spending it all on the research. You need to have grant administrators to make sure the grant is written properly and meets the needs of the funding agency, then you need them afterwards to let you know if you can spend the money you got on the things you want - these grants often times have strict rules on them.

Then there's all the federal regulations. Are you in a lab that got private (not public) money for doing stem cell research? Awesome! Just make sure that any equipment you use (staff payroll, PCs, consumables, anything) hasn't been paid for by a federal grant. So now you have to buy everything twice and make sure you don't cross the streams.

Even if you get a $500,000 grant, anywhere up to 2/3 of that goes immediately to the university you work for for overhead. Aforementioned administrators, physical space, power, cooling, IT...hey, so let's talk about IT for a bit.

So each researcher thinks they're the best thing to ever hit the institution and the way they do things is right. Forget the fact that your IT staff has way more experience and would be happy to help you design whatever you need - they're idiots! So you go off and design your own system and have the grant pay for it, but you ten forget that you don't have any IT staff, so you have a few postdocs take care of it until you realize they're spending all their time working on that and not doing research, so you call up the CIO and yell at him for a while. An IT person shows up and starts identifying problems with your design and why didn't you consult him when you were writing the grant but that's not your concern. So now you're telling the researcher you need a blue Hadoop cluster and you need it right now otherwise you'll take your entire lab across country where their IT staff is apparently more organized than yours. So the IT guy is building the blue Hadoop cluster, burning through IT budget since the CIO promised you they'd take care of it. IT is now underfunded and can't afford the $3 million for a new storage array since every other researcher is doing the exact same thing. But now there's a bigger problem - you ran out of storage space! Where are you supposed to put the 75TB of data you just remembered you needed a postdoc to download? Those stupid IT guys, saying that storage is $.50/GB. I can go to Best Buy and get a 2TB drive for $100! Why can't they just use those drives?

Hmm...I seem to have gone off on a rant. Anyway, a former director described one location as "land of 1000 CIOs". In a way it's true since it's the researchers that are bringing in money, way more than the students. So the researchers generally get their way or else they'll take off elsewhere and take all that research money with them.

And where's my blue Hadoop?

Comment: Re:Not sure what they mean... (Score 5, Interesting) 250

It means that when I deployed a new virtual desktop in Azure and specified "East US" as the data center location, services that looked at the IP address thought I was in Brazil or Germany. Which played hell with Google when I started Chrome because it customized the language for the area it thought I was in. That explains a lot.

Comment: Re:$150,000? (Score 1) 220

by Enry (#47202083) Attached to: NSF Researcher Suspended For Mining Bitcoin

It's probably not arbitrary.

As background, I used to run a cluster at a Major New England University and got involved in some of the chargeback models that were set up. Some of the money for the cluster came from federal funds so I learned some of the high level rules associated with this.

You take all the charges associated with putting a cluster together - the hardware, software licenses, maintenance, system administrators, storage, storage administrators, network hardware and network administrators, data center floor space/power/cooling, pretty much anything that touches the hardware - and you space that out over the expected life of the hardware (usually 3-5 years). As you can imagine, this winds up being a pretty big number. You then divide by the number of cores and CPU hours to get a per-CPU hour number (or you can do wall time since it's fairly rare when jobs are 100% CPU efficient). Once users start using the service, the queueing system can track usage per job and give you some really detailed information on who used what and for how long. If you know this user ran this job for this amount of time, it's not too difficult to figure out how much money was spent.

Why stuff like this is tracked is somewhat important. If federal funds were used to put together a resource like an HPC, federal funds from e.g. a researcher's grant can't be used to pay for access to the resource - the government would be paying twice for the same resource. There are a few ways around it using direct and indirect costs and making large changes to how the resource is set up and managed. In the case of the system I managed it was more trouble than it was worth.

"Marriage is like a cage; one sees the birds outside desperate to get in, and those inside desperate to get out." -- Montaigne

Working...