Linux backup for the rest of us

Enterprise Linux shops have plenty of good options for client/server backup, including Storix, Arkeia, and Computer Associates’ BrightStor ARCserve. But what’s available for a small businessperson running a peer-to-peer network? I have yet to find the equivalent of my favorite Windows backup software, Stomp’s BackUp MyPC. The closest I’ve come — and it’s not close at all — is an application called Konserve. It’s not fully baked yet, but it’s better than nothing.

As you might guess from the name, Konserve runs under KDE. The application has been in development for more than a year and a half, and the current version 0.10.1 was released just last month.

Konserve lacks a main window; it’s designed to run in the system tray. To change its configuration, you click on the Konserve icon. That brings up a menu that includes a Wizard, which helps clueless users set up backup jobs, and Preferences, which lets you add, delete, configure, and activate jobs. Once you have jobs, their names show up at the top of the tray icon menu, and clicking on one activates it immediately.

Konserve’s configuration options are simple. You specify a source URL, a destination URL, and a backup interval, which can be any number of seconds, minutes, hours, or days. When you’re happy with what you have, you click a field to make the profile active and the Apply button to make the changes permanent.

Note that the source and destination are URLs, which means you can specify not only directories and files, but FTP sites as well. I had some trouble with that option, however. My remote FTP server, like any prudent server to which users can write files, requires you to specify a username and password. Konserve uses an anonymous FTP connection, so it failed to complete a job with an FTP destination. Even worse, it failed to tell me it failed. It was more communicative when I tried to create a backup in a local directory to which I had no write privileges; in that case, it reported its failure.

When activated, either manually or when its scheduled time comes, Konserve begins making your hard drive whir. When it stops, you’ll find an archive file in the destination you specified.

The actual backup process worked well for me. I backed up my 477MB home directory, and the software compressed it down to a 342MB GZipped tar archive in just a few minutes.

So Konserve works, but it lacks any refinements that might make it more useful than a tar command in a crontab file. For instance, there’s no option to choose full, incremental, or differential backups — full is your only choice. You can’t pick specific directories or files to include or exclude. Once in progress, you can’t interrupt a backup job. Konserve provides no log file of what it has done. Restore options are non-existent; your only choice is to restore the entire archive, which is annoying when you only need a single file you deleted by accident. When restoring archives Konserve won’t overwrite files in the restore location — you have to remove or rename the original source location.

Still, Konserve works. I’m happy to have my home directory protected now, because I know I’m going to do something stupid someday. (Insert obvious comment here.)

Fun fact: “Konserve” means “can” in German, the native language of the program’s lead developer, Florian Simnacher. Simnacher has adopted a modified Campbell’s soup can as the product’s logo.

Red Hat or SUSE for the enterprise? Hint: Bet the chameleon

Can someone explain to me how Red Hat got to be the most popular Linux distribution in the U.S.? Technology isn’t the answer; other distros have tangible advantages there. If good marketing is the answer, Red Hat’s days at the top are numbered; with Novell behind it, SuSE is going to kick Red Hat’s crown.

I didn’t install Red Hat myself until I’d been a Linux user for many months. I’d already tried more than half a dozen less popular distros: Mandrake 9.0, SuSE 8.2 and 9.0, Xandros 1.1, Mepis 1.1, Knoppix 3.3, LindowsOS 4.0, Slackware 9.0, and probably a few I’m forgetting. At some point I realized I had a crucial gap in my distro experiences. When someone advised me to try Red Hat for a project I was working on, it seemed time to kill two birds with one stone.

I expected Red Hat to be something special. How else, I figured, could it have managed to capture its large share of the market?

Well, it wasn’t Red Hat’s near-universal application support. The range of available applications is adequate, but not nearly as complete as, say, Mepis.

It wasn’t Red Hat’s unique administration tools. There are none. Mandrake‘s group of management utilities and SuSE‘s YaST application are powerful pluses for those distros. Red Hat lacks anything comparable.

Maybe it was Red Hat Package Manager (now RPM Package Manager), the company’s tool for installing binary applications with all the required dependencies in place. RPM is an excellent tool, but it’s no better than Debian’s apt-get application and .DEB files.

Also perhaps a contributing factor was Red Hat’s large roster of resellers, which act as an extended sales and support force. Most distros don’t even dream of having organizations of that scope.

But today’s Linux market is very different from the software bazaar of just two years ago. IBM has upped the visibility of Linux for everyone, and SuSE’s acquisition by Novell has paired a top distro with a top networking company. Many organizations are very comfortable today with Linux on their servers, and may be ready to start considering it for desktops as well.

Microsoft taught us that it’s easier to sell server operating systems when your customers already use your desktop OS. I suspect Novell is going to show us that the converse works too.

With its background as a network operating system vendor, Novell has always been more server-oriented than desktop-focused. But its purchase of Ximian gave it a vested interest in the desktop, and SuSE plays well in both arenas. I expect to see Novell move aggressively into what is today a wide-open market for Intel-based client alternatives to Windows. While many Linux distros would make fine business desktop clients, in reality only a handful of companies are well-capitalized enough to devote the resources to building the kind of marketing, sales, and support organizations businesses expect.

Red Hat already has the organization, but dropping the Red Hat Linux desktop is a questionable tactic. Maybe the company foresaw an expensive battle for the desktop, and maybe that was a big factor in it cutting loose Red Hat Linux in favor of Fedora. (Though why they picked the name of an existing open source project is another mystery.) A company without both a compelling server and desktop product is bound to be at a disadvantage when trying to sell to the enterprise. No, Red Hat Linux is nothing special, but if one man can code a top-notch distro like Mepis, surely Red Hat’s development staff could do as well or better.

Of course, each business selects what it feels is the best distribution based on its unique problems and needs. But given today’s market, I can’t see any situation where Red Hat and Fedora are likely to be better than the competition.

That doesn’t mean Red Hat can’t continue to win market share. We’ve seen plenty of examples of inferior products that sold like Pet Rocks because of good marketing. But when I think of good marketing in the computer industry, Red Hat is not a name that immediately stands out.

Luckily for Red Hat, neither is Novell. That company could have had the network operating system market locked up 10 years ago if it had been able to sell the benefits of its Novell Directory Services. Instead, Microsoft introduced the inferior Windows NT Server, and through a combination of superb marketing and questionable business tactics pushed Novell to the brink.

Novell’s technology is still first-rate, and its acquisition of SuSE and rapid adoption of open source development methodologies is giving the company new life and energy. Today, SuSE is number two in U.S. market share, and its prospects are excellent, thanks both to the added value of Novell’s network services, which are being ported to Linux, and its new parent’s commitment to regaining enterprise mindshare. By contrast, Fedora is a cast-off project from a company without significant technology differentiators.

The technology marketplace is always volatile. Today Red Hat is the default choice for U.S. businesses. Two years from now, will it even be in the top three?