April 29, 2009, 9:50 AM — Novell released SUSE Linux Enterprise Desktop 11 (SLED 11) at the same time it rolled out SLES 11.
We found that the desktop bundle (which we tested on an HP EliteBook 2530p notebook system) includes features that render a look and feel similar to past versions of Apple's MacOS as well as some parts of Windows 7 beta we've tested.
SLED includes a lot of updated software including the latest Gnome desktop and all that it includes. There are proprietary drivers for ATI and Nvidia graphics cards available for the desktop version. The latest version of FireFox with Flash, Silverlight 1.0 (via Moonlight) and Java support, the Banshee media player, the F-Spot photo browser, OpenOffice, the Pidgin IM client, and the Evolution calendaring and e-mail program are all included.
Unlike other desktop-focused Linux distributions, SLED isn't supplied with the libraries and 'guts' needed to compile other downloaded applications. Although this can be remedied by gathering the components from various places, that process can be painful. Average business users may not care, but this applications lock down might make the Linux cognoscenti boil.
Running with a Gnome-based user interface connected to Compiz (a compositing window manager X Windows which uses 3-D graphics hardware to create desktop effects for window management), we saw some interesting eye candy that features quick animations and visuals, like shrinking windows.
SLED's dual listing of certain applications in both the YaST resource management window and in the Control Panel had us scratching our heads on what to pick out of the many open source applications included with the bundle. This confusion extends to having Hardware>Control Center choices duplicated as Network Devices under YaST as well. There's no real harm in this double listing, it's just confusing.
Beagle is an indexing service used by SLED 11 and it's similar to MacOS X's Spotlight feature and Windows 7 indexing. Beagle was quite flexible in our testing, allowing us confine our indexing to appropriate areas, rather than the entire system, for speed-of-search purposes.
Wireless attachments to our IEEE 802.11b/n router from Linksys worked flawlessly, as did IPv4 and IPv6 networking. But not all the features of SLED 11 performed with aplomb in our testing.
SLED 11 was not able to find all the system hardware, although it did a very respectable job. For example, there was a fingerprint reader installed on the HP EliteBook 2530p, but it was ignored by hardware discovery and no device drivers could be found. There was no network printer discovery protocol employed, so we had to enter printer IP addresses manually to make our connections.
We tried to get Bluetooth working -- and it did -- but only partially with our cell phones, as several OBEX (Objext Exchange) functions aren't available.
SLED 11 starts to approach the slick look and feel of competitors MacOS 10.5 and Windows 7. The applications supplied will be easily accessible to most, and, like Linux competitor Ubuntu, SLED 11 has the Mono Framework for .Net compatibility. Could we give this to business users and make them productive? Yes. And make them happy? Perhaps.
How we tested SLES, SLED 11
We tested SLES and SLED 11 on a switched Gigabit Ethernet network (Dlink managed switches). We tested SLES 11 principally on an HP DL580G5 server (with four sockets, 16 64-bit cores, 16GB memory, internal SAS array, 4-Gigabit Ethernet connections).
We tested connectivity between the various servers mentioned above and clients (Windows XP SP2, Windows Vista, and Windows Beta 7, as well as MacOS 10.5.2, and Ubuntu 8.04.), installation and manageability.
Performance was tested via a regimen we developed for comparing native and VM guest server performance using SPEC's SPECjbb2005, and Intel's IOMeter.
We tested SLES 11 Xen performance using SPECjbb2005, a Java-based business simulation test, with three SLES 11 inside VMs running Java 1.6 comparing with SLES 10 SP2 VMs running Java 1.6. The second test was to test SLES 10SP2 under SLES 11 Xen against SLES 10 SP2 under SLES 10 Xen, both of these used Java 1.5 for comparative purposes.
We also used IOmeter to measure the native SLES 11 IOs/second vs. native SLES 10SP2 IOs/second in terms of disk performance.