Chrome

2009-11-24 09:55 am
mdlbear: (hacker glider)

Today's feeds included this post on ChromeOS by Cringely.

Last week Google made a preemptive strike against Microsoft, revealing details of its Chrome OS months before that product reaches its near-infinite beta release. The idea is simple: who needs a big OS if you are doing everything in a browser? It’s a huge threat to Microsoft and Apple. But then it struck me I’ve heard this all before, so I went back and found this video clip from my show Triumph of the Nerds, circa 1996, where Larry Ellison predicts the future, not knowing he was actually describing 2010.

It struck me that I've seen it before, too: Sun's Display Postscript. The difference is that Javascript is a far friendlier language to program in than Postscript, and that Chrome has some solid security behind it. We've learned a lot in the last two decades: Javascript was designed from the ground up for implementing the programming side of a user interface, and HTML 5 plus SVG on the front end are more than good enough to replace a 2-d graphics toolkit like GTK. So there's that.

Sun also invented the catch-phrase "the network is the computer", and NFS (which stands for Network File System, in case you've forgotten).

Cringe ends with:

We know that under the Chrome OS Google Apps will be very secure. Any tampering will trigger the download of a new and pure OS image. But will the Chrome OS have enough performance to compete with Microsoft Office? I think it eventually will, based, for example, on extensions like Google’s recently announced O3D API, which will allow Google Apps and approved third-party apps to grab spare GPU cycles to improve performance.

What’s left to be seen here is whether these improvements will be enough to beat Office or if Google will have to make a standalone (local PC-based) version of these apps. Only time will tell.

The most interesting part for me will be Microsoft’s response. This strikes at the very heart of Redmond’s business success and Microsoft will not take it lying down. Expect thermonuclear warfare.

What he seems to be forgetting is that it's going to be a three-front war. Down underneath ChromeOS, Android, Maemo, Moblin, and the Ubuntu Netbook Remix -- all of which are designed for mobile devices and thin clients -- is a perfectly functional Linux kernel. Mass storage is still dirt-cheap, and even after we get fiber to the home a suitcase full of hard drives will have more bandwidth than a pipeline to the cloud.

And a net app doesn't much care whether the back-end server it's talking to is in the next room or the next state, but you'll notice the performance difference. Heck, I notice the performance difference in my favorite audio editor between a local SATA drive and the same drive over gigabit ethernet.

Yeah, you'll back up to the cloud, use it for communication and sharing, and use it for deploying massive multi-user web applications. But all your applications will be running their front ends locally on Linux, so they can sneak behind the web to the local filesystem, and sneak behind the browser to X and the GPU, when you need performance. Your photos and music and video may be synchronized to a big server in the sky, but they'll all be in the fileserver in the closet, too. And some things won't get synced; the reasons are left as an exercise for the reader.

No matter what the (other) pundits are trying to claim, a URL is still a (virtual) pathname, a website is still a hierarchical filesystem (which you can of course mount with WebDAV), and the hierarchy of INTERconnected NETworks doesn't stop at your DSL modem or even at the ethernet port on the back of your PC. PCI is a network, too.

ChromeOS may take down Microsoft, and the Google Store may become a strong competitor for Apple. But is it the end of operating systems as we know them? I doubt it. Will you get all your applications from the Google Store? Probably not, any more than you'll get all your music from iTunes and all your books from Amazon. You'll hit some indy website, download to your local server, and away you go.

mdlbear: (kill bill)

In a word, "no". At least, not to anyone except the obvious competitor. But I'll get to that later.

Much ink and many pixels have been spilled recently over Oracle's announcement that it will be providing "their own" Linux distribution. In particular, what they'll be doing is exactly what CentOS and White Box do: download a stack of sources from Red Hat, rebuild them without the branding package, and call it "Oracle Linux".

Here are a few articles from InfoWorld that I came across this morning: "Oracle to push Red Hat from support chair" (yes, Oracle will be providing their own, paid support), "Oracle-Ubuntu rumor fizzles" (a lot of observers expected Oracle to simply certify the Ubuntu distribution), "Oracle move a worry for Red Hat" (um... yes; maybe), "With friends like these....(Oracle goes after Red Hat)", and finally this one pointing off to this entry in Dave Dargo's blog at Ingres. Dave "was a longtime Oracle employee and started and ran Oracle's open source program office", and now works for one of Oracle's open-source competitors.

Dave Dargo is understandably skeptical of Oracle's ability to compete with RedHat on the basis of support: "There's a survey from CIOInsight [PDF] that shows Red Hat is the number one vendor for value as rated by CIOs in 2004 and 2005. Where does Oracle fit on that chart? Glad you asked, they ranked 39 out of 41." But nobody seems to be asking themselves who Oracle's real target is.

I'll give you a hint: it isn't RedHat. RedHat has a solid number-one position in enterprise Linux, a solid number-one reputation for support (even if it's a little slow sometimes), and it's really not very likely that any of their customers are going to switch, unless they already want to run an Oracle database server. If they do, they'll be happy to get their OS and their database from the same vendor, and pay the additional 5% for the convenience of getting all their support from the same place. What they get with that support is a guaranteed lack of fingerpointing. As somebody who once bought a RedHat-based server from Dell, I can assure you that that's well worth the price.

So who else makes a big marketing noise about being a one-stop shop for your enterprise database server? Microsoft, of course. That's Oracle's real target. Microsoft, who can tweak their database server and their OS kernel to work perfectly together -- or at least get in their competitors' way when it comes down to running benchmarks. Now Oracle can play that game. Until now, Oracle couldn't offer an "all your software from one vendor, guaranteed no fingerpointing, just runs out of the box" solution. Now they can. It's that simple.

I expect that the impact on RedHat's non-Oracle customers will be minimal. They'll be hit a little more than other distros, because almost all of Oracle's customers were already running RHEL. Maybe they'll be hit a lot more; it's not clear how many of their enterprise customers weren't running Oracle. But RedHat isn't number one anymore, and having multiple vendors in the enterprise (high rent) market can only benefit the Linux market as a whole. Who knows? Maybe it'll convince RH that they should go back to selling $99 boxed sets like they used to do. Because if they don't, Oracle will get all that lovely shelf space in Fry's.

But the hackers will stick to Debian, Ubuntu, Gentoo, Fedora, and Slackware, and probably won't even notice. The IT guys who just want a box at home with the same style of config files as their big servers at work might download Oracle instead of CentOS or White Box. Or might not. Not a problem. RedHat will lose some customers, but if they go back to selling boxed sets at retail they won't lose much, if any, off their bottom line. The big loser, at least if everything goes according to Oracle's plan, will be Microsoft.

And that's as it should be.

Most Popular Tags

Syndicate

RSS Atom

Style Credit

Page generated 2025-07-27 12:29 am
Powered by Dreamwidth Studios