CES came, CES went, and like every year, I ignored it. It might be the size of my apartment, or my checking account, but I don’t have a burning desire to own a TV that’s taller than me. Or to hook it up to a media center that can fit in my pocket. Nor am I fascinated by cell phones with mediocre reception in four different bands! And I already settled the Blu-Ray vs HD-DVD debate by choosing neither one.
So excuse me for not paying attention. But there was one piece of seriously interesting news. And, ironically, it has less to do with consumer electronics. SanDisk started selling Solid State Disks.
Solid State Disks (SSD) don’t mind when your notebook loses balance and hits the floor. But even better, they eat up less battery when running, and can quickly suspend to disk instead of suspending to RAM. So what do you do with all this new found battery charge? Pipe it to the radio transmitters: WiFi, Bluetooth, Cellular, WiMax.
We’re getting one step closer to always-on connectivity, on the desktop and the notebook. It will happen first in executive toys, which explains why SanDisk chose CES to bring us the news.
When SSD go mainstream, they’ll change the way we use computers. You trade local storage – 32GB is not a lot of space for a modern operating system, music collection and the occasional ripped DVD – for the ability to live in the cloud: to always have access to more information that any one computer can store.
And, it will change the way your software runs. When there’s little penalty in saving to disk, there’s no reason not to. Your changes are stored as you’re writing the memo, or playing with the spreadsheet. I won’t shed a tear over the loss of the Save button.
We can almost do it today, we have auto-save that creates a backup in the background every minute or so. But if you crash, you lose your place in the document, the last few keystrokes and all your undo actions. So you still need to commit to a change, to let the software know you’re ready for what may come. That will no longer be necessary.
But what I’m most excited about is the way it changes our computation ability.
Every year, like clockwork, hard disks get bigger, faster and cheaper. Yet, we’re still hitting the same wall, the slow speed by which the needle dances from track to track, waiting for the right sector to pass underneath. Relatively, the bigger the hard disk, the slower it can access the data. That’s why we call hard disks the new tape.
With SSD, the hard disk will literally become the new tape.
SSD has constant seek time, so it doesn’t matter where the data resides. It’s as efficient streaming a movie as it is collecting records scattered all over the database. SanDisk claims it’s a hundred times faster, but even an order of magnitude is enough to change the rules of the game.
All of a sudden we can build applications that are smarter by virtue of having more data on tap. We can be more liberal with our reads and writes, expand our queries, use more meta-data.
It all comes from the simple fact that hardware has to pay for itself. $18 affords you 1GB of SSD (expected to halve every 18 months). You’ll have to pay nearly ten times as much for RAM. Of course, both are considerably more expensive than magnetic media, so video and backup are not the killer-apps for SSD. But video and backup work fine with tapes.
Think of the way we develop software for servers. You can only afford so much RAM, that’s your upper limit on the amount of data you can process quickly. Disk access is so slow, it’s the first bottleneck you optimize around. You work around the snail-store by sticking to the confines of affordable RAM. You may have 250GB of storage in there, but you’re building the software to live in 4GB or less.
What if we opened the floodgates? What if you had 2GB of RAM to compute, 32GB of SSD for fast random access, and 250GB of the slow kind. How would that change the way you design, and the kind of features you build?
Now granted, hardware always improves at least at constant rate, and software expands at the same rate to fill it up. But sudden increases lead to sudden expansion, and sudden expansion comes not from incremental improvements, but innovating new ways we couldn’t afford before.
Think of what happened when storage became small and cheap enough to fit your music collection in your shirt pocket? Or when bandwidth jumped from fast enough for text to fast enough for video?
This change will mostly affect the dark code, the plumbing that runs behind the scene, but out of that will come new types of applications we couldn’t imagine a year ago. Or at least I couldn’t, but I’m starting to have some ideas. What would you do with 32GB of fast, durable memory?
