Follow Slashdot stories on Twitter


Forgot your password?

Comment Re:400 Mb per seconds (Score 1) 60

Not really. The real-time components (aka correlation) are basically just straight up FFTs. Custom hardware in correlators might make sense (and probably does at scale), but through ASICs or FPGAs. They're not doing that (...yet). Throw a GPU or two into each node, and you'd get far more FLOPS than you would with a cray. This work is mostly embarrassingly parallel, so throwing money into cray's is a total waste of time.

Comment Re:400 Mb per seconds (Score 1) 60

It's not. It's really not much at all. For $150k, you could build a hadoop cluster that would happily accept the data stream, process it, and make it available for consumption. If you just want to store it, you don't even need that much.

That's a waste of a Cray. Well, a Cray is a waste of money these days anyway.

Comment Re:Not much of a sample size. (Score 5, Insightful) 224

I actually thought this through... my first reaction was wow, that sucks. Maybe Valve isn't the utopia that people think it is.

Then I stepped back and remembered what I've heard about Valve. You make your own decisions - and you're accountable for them. They said they had a million dollar lab, but couldn't hire anyone to do the machining. But who decided to build that lab? Did they spend a million dollars on equipment then not use it?

A flat organizational structure doesn't mean there's no politics. It means politics are MORE important - it's harder for some team to simply burn cash, because everyone's eyes are on you. It's hugely increased freedom - but all of the responsibility that comes with it. Assuming that anyone in Valve could decide to go build a million dollar lab, what do you think would happen if it failed to get utilized?

This is one side of the story from one person. I'm sure there's more to it than the lab, but the lab example shows a basic misunderstanding of the personal responsibility one has in a flat org structure.

Comment Re:Innovation (Score 4, Interesting) 257

Durability is not generally designed into current smartphones. Intentionally. Think about the volume of sales, and how they would decrease, if devices didn't break within about 2-3 years. In fact, durability nearly sank RIM - as most people were comparing new iPhones with the original bold - released at the same time as the original iPhone.

2-3 years is an eon in mobile right now. If devices don't die, people don't upgrade.

Comment Re:Does it compute? (Score 2) 168

Except for a few problems:

Hadoop tops out around 4000-6000 nodes, then you run into serious scalability issues in the jobtracker and HDFS scalability. Granted, with HDFS federation and YARN these should improve, but today you can't build this wider than a few racks without spending a good chunk of time doing some significant hadoop engineering.

Second, disk. Where's the disk? Hadoop needs disk. Hadoop likes disk. Disk likes hadoop. Hadoop likes lots and lots of disk. Nice, you've built a 6 watt SoC. Now just put six 10 watt 1TB drives beside it and I'll be happy. Oh, and make sure the disk controller can do spindle speed on all six.

Comment Re:how do you get the Cs to "get" it? (Score 2) 201

In my case, I didn't even talk about the major advantages like the complex analysis. 100 billion row joins don't sell execs. I did it purely based on the cost vs. enterprise grade storage. That got it in the door - and then I was selling the platform to developers. Showing a dev team a little pig script written on the spot, then sending it out on a production cluster and watching it use 800 cores and 8TB of memory, processing a few dozen TB of data while we're sitting in a room, and the devs got on board. Now I've got new dev groups wanting onto the platform regularly - and the execs keep hearing about how this hadoop thing is a critical part of their app/service/report. It sells itself, after a point.

Comment Re:"Big Data" (Score 5, Insightful) 201

I'm biased, as my entire job is building those systems so many people refer to as "big data" - but the marketing is terrible. The technology itself is quite good, and makes a huge amount of sense. The problem is, companies traditionally used to doing data stuff for large corporations (ie, EMC, Oracle) are pissing themselves. This destroys their entire business model - so they're flooding the market with crap trying to avoid losing absolute boatloads of money and accounts to these technologies.

Talk about big data with those companies all you like, and they won't mention the actual reason hadoop and the like are a big deal:
1- It's all open source. Don't wanna pay? Self support.
2- It's all designed to run on the cheapest commodity hardware you can find. Why buy appliances with huge markups?

This has companies used to huge margins on appliances and software shitting themselves. They tried FUD with single point of failure stuff, and now that that's solved, they're stuffing infiniband into custom rack designs and saying how much better it is. Meanwhile you can buy 4x the gear for that same price.

Is it a buzzword? Yup. Is it saturated with marketing? Yup. Is it a stupid idea? Hell no.

Comment Re:Looking around me... (Score 5, Insightful) 189

I actually had the opposite reaction. I'm 31, and I constantly fight the urge to run everywhere. I remember all through school, even into early highschool, I'd run everywhere I wanted to go.

Then it was uncool to run. Then inappropriate. Then unprofessional. A year ago, effectively 15 years after I stopped running everywhere, I started running for exercise. I'm getting back into shape. And I'm finding it annoying that I can't just run all the time - I'll get sweaty or smelly, and that's just socially unacceptable.

I'm pretty sure we're all meant to run a LOT more than we do - and we've forced ourselves to stop due to social pressure.

Comment Re:Idiot. (Score 1) 633

You're an idiot. You signed something under threat of prison / arrest without bothering to consult a lawyer.

Actually, that's exactly what he should have done. If you're ever presented paperwork, and presented a choice at which point you must choose right there and then, sign it. It's compelled/under duress.

Several of my former employers had a policy on termination where you must sign an agreement stating you won't sue for anything (wrongful dismissal, etc) in order to get a generous severance package. But, you have to make that agreement there, in the room, before they walk you out of the building. I ran this past a few lawyers, and they all said the exact same thing: if you're under duress, sign it right away. Agreements that are signed under duress are void.

Slashdot Top Deals

It's fabulous! We haven't seen anything like it in the last half an hour! -- Macy's