I had to replace my old LAMP web server due to a hard drive failure so I decided to give Ubuntu Feisty Fawn server edition a whirl. So far I’m pleasantly surprised. I had some requirements going in:
LAMP Stack – Needs to be an up to date, Apahce/PHP/MySQL based web server
Webmin – Because I feel emasculated when I plug a mouse into a server but the command line slows me down sometimes
PHPMyAdmin – It just makes life easier
SSH – Though the server is on a laptop with an LCD, etc., I prefer to config from other machinees
SAMBA – I develop with Notepad++, which I can’t get running in Linux/WINE, so I wanted to map a network drive, though BlueFish looks interesting
PEAR – Why reinvent the wheel if you’re developing PHP web apps?
Setup was a piece of cake for the most part. If I ever find time I’ll put together a howto combining all of the steps involved in building this thing. Webmin, PHPMyAdmin, SSH, SAMBA and PEAR should all be options in the installer tool at some point to make our lives easier.
As for the desktop version, it’s a huge improvement, at least on my Dell Latitude D600 Laptop. My resolution was correctly set without intervention (thank god – Xorg.conf is evil) and flash for Firefox was a piece of cake.
Wireless networking was the only thing that didn’t work. Because I’m a bit of a nerd I figured out how to get it working but it took some research. In case you’re curious, plug in a regular LAN cable to get online. Go to System > Administration > Synaptic Package Manager and click Search. Type in fwcutter and install the tool it finds. Then check Fetch and extract firmware and click Forward. Reboot and voilà, works for many of the Broadcom Truemobile cards.
This ease of installation is really remarkable considering the headache of an XP or Vista fresh install. I had one minor hurdle and now I’m one step closer to an entirely open source existence. Not that I’m opposed to proprietary but I prefer not to spend money if at all possible. Oh and whatever you do, do NOT turn on the desktop effects if you’re using a laptop with a ATI Mobility Radeon 9000.
Ed. Looking for time to clean this up but some good info in the interim.
My server hard drive just died, I fixed the corrupted registry with the Ultimate Boot CD 4 but it’s a losing battle, too many moving parts. But this annoying little episode got me thinking. The hard drive business is huge ($30Billion anually) but a new breed of Solid State Drives (SSDs) with no moving parts are starting to crop up and compete. With prices on the flash used in these new drives plummeting combined with diminishing returns on storage growth (does your average consumer really need a 1 terrabyte hard drive?) plus the promise of better performance and lower power consumption, the industry as it exists today is staring down a capitalist abyss. Schumpeter called it Creative Destruction. I call it better stuff cheaper.
Cost Per Gigabyte:
Decent DDR2 RAM is going for about $49 per Gigabyte on NewEgg.com (as of June 25, 2007)
Flash based drives are around $16 per Gigabyte (based on the Samsung)
Standard drives with moving parts range from $.30 to $1.20 per GB depending on size and use (laptop drives) Lets say $.50 for simplicity.
Again, I think it’s important to keep in mind that most people would probably be happy with 100GB of storage. So even though the cost per Gigabyte is low on some of the new jumbo drives the total cost can be higher than for a small flash based drive.
Performance:
RAM Drives will be bottlenecked by their connection to the computer. So if you have SATA1 connection on your motherboard that’s 150MBytes/s (1200Mbits/s = 1.2Gbits/s) which is a tad faster than your Gigabit LAN connection). SATA2 is 300Mbytes/s (2.4Gbits) and PATA topped out at 133Mbytes/s though that would clog a PCI bus. Another thing to keep in mind is that you can write just as fast as you can read data from a RAM drive as you’re still bottlenecked by your connection and not the drive (in theory, actual speeds on the RAM Drives are closer to 140MB/s Read and 130MB/s Write as of now).
Here’s a video to give you a sense of the performance you can expect from an I-RAM.
Flash Drives are typically faster than regular hard drives in overall performance but lag behind when you’re doing large file copies. The lack of moving parts also cuts down on power consumption which is great for laptops. Laptops also suffer from slower drives so some flash based SSDs do beat regular laptop drives in all performance benchmarks. The Samsung drive reads at 50MB/s and writes at 30MB/s. Compare that to a RAM Drive’s 150MByte/s read and 150 write on a SATA bus. SATA based flash drives won’t see much of a performance boost compared to the current PATA offerings but it would be nice to have them around for newer PCs and laptops. If flash based drives can hit 150MB/s they would appear to be just as fast as RAM drives assuming you’re using SATA1, even using inferior technology.
Here is a video of SanDisk’s SSD vs a typical laptop drive. 35 seconds vs 55 seconds. Fast but nowhere near what the I-RAM can do.
Size and Practicality:
Laptop Flash Drives: 16 Gigabytes is my minimum drive size for a laptop. With that I can install Windows XP (or Ubuntu) and have room for Office, Itunes, and still have room for some songs and videos. 8Gb could be doable if I just needed a Office and web browsing. And wouldn’t ya know it most of the 8GB SSD Flash drives are sold out on Newegg.com. Their out-of-stock 16GB version is going for $258 and the in-stock 8GB model is selling for $142.
Laptop RAM Drives: Unlike svelte 2.5 inch flash drives, huge sticks of RAM are never going to work in laptop hard drives. 2GB, SODIMMS (Laptop RAM) are smaller and might fit but limited battery life means RAM drives in laptops are probably a long way off.
Desktop RAM Drives: You don’t want to tote around an external drive if you’re using a laptop but at home it makes a lot more sense to have a super fast system drive that boots in 5 seconds paired with a cheap 500GB monster to large videos, MP3s and the like. For some unknown reason, Gigabyte hasn’t released an upgraded I-RAM in spite of the overwhelming demand. The successor simply adds a different form factor so you don’t need a PCI slot to use it but the size and performance haven’t improved in well over a year, odd for the PC industry (conspiracy theory pending).
Based on the numerous discussion threads about the I-RAM people want something with the following specs:
Support for 8Gigabytes+ per drive.
DDR2 support because it’s the best bang for buck at this point.
ECC to make sure nothing goes haywire.
External drive support (ESATA) which would allow people to use their own Un-interruptible Power Supplies.
This device does exist in the form of the HyperDrive4. The price? $2,325 + ECC RAM at about $100 per GB. A 16GB Hyperdrive storage system would set you back about $5,000. If Gigabyte could put together a cheap card with 8 slots and SATA2 support for less than $1000 they’d have a winner. A 16GB I-RAM3 would then cost about $1000+$800(16GB DDR2 RAM) or less than $2000 for a system with better performance than the Hyperdrive which is stuck with SATA1. The era of the hard-drive as a bottleneck in performance would end.
As CPUs and other computer components outstrip the average user’s need it makes sense to focus on bottlenecks and reliablility. The hard drive as we know it is the slowest, most unreliable, and power hungry component in a modern system.
I see three markets for hard drives.
The Performance Enthusiast (Mine is faster)
The Media Storage Buff (Mine is bigger)
Joe Six Pack (My minivan needs new tires)
The performance category could be wiped out by an improved I-RAM and the JSP market could be dominated by a relatively fast, reasonably sized flash drive when prices inevitably drop. That leaves a $30 Billion industry wondering how to keep shareholders happy.
I signed up as a tester long ago but finally got the invite. This is some really cool technology. (Edit: After using it a bit, I’m convinced this will be gigantic).
They’ve basically created something like Wikipedia but instead of being limited to documents you get to create your own objects (called types, see image). For instance, a film type would consist of a director, actors, rating, etc. Once created, anybody can update the data ala Wikipedia. Wikipedia has this capability but it’s not nearly as structured. For instance, I can search for Scorsese on Wikipedia but on Freebase I can search for Scorcese films released in the ’80s featuring Robert Deniro. Or all Porsche models released between 1967 and 1971 with engines greater than 3 liters. It has the structure that Wikipedia has been lacking.
This is the promise of the semantic web. A database of structured searchable knowledge. In theory you could write software to look for patterns in movies which could make suggestions. For instance. If Deniro is in a bunch of Scorsese movies and a bunch of Speilberg movies a recommendation system would know you like Scorsese and suggest Speilberg movies. Not even remotely possible with Wikipedia.
If they can create a quick way to import Wikipedia data they could crush the competition in short order (update, looks like this is what they’re doing “The Film domain is one of the most complete domains in Freebase with nearly 23,000 films, about 9,700 of which have structured information loaded from Wikipedia infoboxes.”). Wikipedia is an encyclopedia. Freebase could be an encyclopedia but also a replacement for IMDB, the Census Bureau and just about any other data source you can imagine. I just updated San Diego County and it works like Wikipedia but with much better use of AJAX.
The really interesting thing is that you can use their database in your applications. So I can write a front end application to navigate through movies on my computer using data from Freebase. Then if someone finds cover art for an old movie residing on my PC it would automatically show up in my application.
I can’t say a heck of a lot about what I’m working on because of two things. One, the last time I put one of my business ideas on this blog it was quoted by media critic and journalist Jeff Jarvis and a flood of curious traffic from Tribune.com killed any chance I had to spring my idea onto an unsuspecting world. Luckily that wasn’t my only idea.
Reason two is that if you put the words “business idea” in your blog, your idea will be found by the dozen or so people who scan blogs looking to harvest ideas free of charge. It’s interesting to see where these people come from. I’ll update this post with that info if I get any nibbles from idea harvesters. If you think your bright idea will remain hidden, even on an unknown blog, think again.
My idea is a tad ambitious and that means it takes time to get the software running well not to mention marketing for a launch and all of the various things that need to happen before you see your first revenue. So I’m taking my idea and applying it to a different market which should get some revenue flowing. I went in to this job with a very clear idea of exactly what I wanted to do but my idea has morphed for the better with some guidance from people with more experience starting companies. I made an assumption about who could be a customer but stepping back it’s clear that the idea is big enough that it can be applied to various other markets.
If my credit cards were maxed out I couldn’t have attempted to start this thing so my advice to fledgling entrepreneurs is to live like a startup founder before you actually start the thing. The funny thing is I’m eating a lot better now that I’m forced to manage expenses. I still have health insurance so I figure the worst that can happen is that I end up with some great experience and the knowledge that I won’t be old and have to wonder what could have been.
Image is called Working Jack by Marc L’esperance. Read about it here.
Don’t have time to invent this but my computer experience would be vastly improved if I could control the volume on my computer using my mouse’s scroll wheel. Maybe if you held down the right button while scrolling it could change volume. I’m really tired of looking for the volume widget on the various applications I use and the volume knob on my speakers is poorly designed.
EDIT: Apparently not a totally novel idea. You can download an application that does just that called Volumouse. It’s free. Gotta love the interweb. This functionality really should be built into operating systems.
My new job doesn’t have defined hours. When I’m not working I feel guilty so time has been short lately. Lots to write about though. Ron Paul is kicking some butt in terms of fund raising and online support. The CNN Political Ticker wrote briefly about the Ron Paul Phenomenon. The story prompted 1,349 comments from Ron Paul supporters concerned about mainstream/mass media bias. I’ve never seen anything like it.
I’ve been following Ron Paul for over a year and generally agree with his points, certainly more than any other candidate, but he seems to lack a bigger picture view the world. EDIT: I was wrong. A reader pointed out her take on this subject in the comments. For example, were the gold standard to be adopted today the economy would probably collapse in a deflationary spiral. Austrian economics looks to be spot on but the global economy isn’t a blank slate. I’d like to hear how he’d manage a transition to a gold standard without causing economic mayhem.
About my new job. Artificial Intelligence Fellowship. No doubt the weirdest job title I’ll ever have, should prompt some guffaws if I put it on my resume. When I’m not writing code I try to stay up to date on the changes taking place in the media and journalism arenas. Two great articles here and here. The newspaper is dying and my project is the blue toothed vulture hovering overhead. You can learn a lot about a country from that vantage point.
I went to Cal Tech and interviewed a handful of potential interns. Their career center gave us free hand-cranked flash lights and were incredibly helpful. A quote from one of the interviewee’s resumes under publications: “Decomposition-based Optimization Algorithm for Parallel Machine Scheduling Problem”. Needless to say I stuck to the less technical interview questions.
I had to replace my old LAMP web server due to a hard drive failure so I decided to give Ubuntu Feisty Fawn server edition a whirl. So far I’m pleasantly surprised. I had some requirements going in:
Wireless networking was the only thing that didn’t work. Because I’m a bit of a nerd I figured out how to get it working but it took some research. In case you’re curious, plug in a regular LAN cable to get online. Go to System > Administration > Synaptic Package Manager and click Search. Type in fwcutter and install the tool it finds. Then check Fetch and extract firmware and click Forward. Reboot and voilà, works for many of the Broadcom Truemobile cards.
I signed up as a tester long ago but finally got the invite. This is some really cool technology. (Edit: After using it a bit, I’m convinced this will be gigantic).
I can’t say a heck of a lot about what I’m working on because of two things. One, the last time I put one of my business ideas on this blog it was 
Don’t have time to invent this but my computer experience would be vastly improved if I could control the volume on my computer using my mouse’s scroll wheel. Maybe if you held down the right button while scrolling it could change volume. I’m really tired of looking for the volume widget on the various applications I use and the volume knob on my speakers is poorly designed.