Choosing the right Virtual OS: Windows XP vs. Windows FLP
One of my favourite things about the ubiquity of multi core CPUs, broadband and gigabytes of RAM in modern computers is that virtual operating systems are finally viable. When I was growing up, trying out another OS meant multiple hard drives, partitions and the struggle of actually tracking down discs for these exotic digital curios. (I learned the hard way about making sure you’d partitioned a drive properly when I overwrote Windows 3.1 with a copy of BeOS I got with a computer magazine sometime in the mid-90s.) Thankfully, broadband can hook you up with pretty much any *nix OS in 10 minutes flat these days, and it’s one of my favourite modern software toys which gives you a no-risk sandbox in which to experience the object of your curiosity: the virtual machine.
I wrote a post at the beginning of this year detailing how to set up a remotely accessible cloud-y virtual machine using VirtualBox 4. Covered in that tutorial were the technical instructions for actually setting up the virtual machine itself, however I didn’t go into any detail about choosing an operating system to run on your brand new VBox (mainly because I specifically needed an XP machine at the time).
To serve as an interesting comparison, and also to segue nicely into my forthcoming (I promise!) series on the ‘OS less travelled’, this post will cover the differences in functionality and performance between ‘full fat’ Windows XP Professional and ‘semi-skimmed’ Windows FLP.
Windows FLP stands for ‘Fundamentals for Legacy PCs’, and is a slimmed down install of XP specifically designed for low-spec machines. You can read up on the OS itself at Wikipedia – but is it a viable candidate for a general purpose VM?
Time to get (relatively) scientific. The two machines will be set up as per my guide with identical specs: 512MB RAM, single CPU, 12GB hard drive and 8MB VRAM.
(Handy aside: to get the specs of a VirtualBox VM, type ‘VBoxManage showvminfo XP’, where XP is the name of the VM in question, at the command line)
The installer for FLP is very different from XP. At the time of its release in 2006, Vista was about to ‘replace’ XP as the computing world’s de facto operating environment. Ironically, the design language of Vista, derided for its over-emphasis on power-sucking visual trinketry, has infiltrated this ‘low power’ Microsoft OS – you can see the modern Windows flag in figure 1, for example. The installer also has a full GUI, a pioneering feature for a Microsoft OS.
Figure 2 shows the installation type selection – I’m going for ‘typical’ in this case. The fully-GUI installation procedure allows very easy access to advanced installation options such as unattended installations and manual TCP/IP configuration. The ‘configure first, install later’ approach is fantastic. All of these ‘pro’ features add up to the best Microsoft install procedure I’ve ever experienced – see figure 3 for unattended and RDC options.
For speed, ease and ‘pro’ attitude, FLP definitely has the superior installer.
Footprint & Desktop
Comparing the footprint of a full-featured OS with a system specifically conceived as a slimmed down version produces not particularly surprising results. The base install of FLP requires just 700MB, as opposed the 3GB+ demanded by XP Professional – whether the system is running locally or ‘in the cloud’, a saving in drive space is always a good thing.
Retained in FLP are the visual effects – the default green/blue theme of XP and the olive/silver alternatives are still available, something I found surprising given that I’ve never encountered an academic or corporate environment which used anything other than Windows Classic.
A small footprint was one of the things FLP was designed for, so it isn’t surprising to find the base install to be less than 25% the size of XP
With 10 years as the world’s most common OS, Windows XP has probably the broadest software catalog of any platform out there. Anyone who’s ever used a hacked ‘lightweight’ XP system will tell you, however, that when you chop bits of the operating system off you often kill compatibility. Of course, these hacked XP installs are just that - amateur versions of the in-house chop-job Microsoft has performed to produce FLP – so how does their ‘pro’ version stack up?
I tested FLP with an identical software set to my XP Professional install, including common packages such as Firefox, .NET 2.0, Office applications and WinRAR, and experienced absolutely no compatibility issues. The Oracle VirtualBox Guest Additions CD installed the Windows guest extensions and drivers with no hiccup, adding shared folders, resolution changes and reducing host CPU load.
(Handy aside: to add a share to a VirtualBox machine, enter ‘VBoxManage sharedfolder add XP –name windowsShare –hostpath /home/windowsShare’ at the command line, where ‘XP’ is the name of your VM and ‘/home/windowsShare’ is the path on the host machine)
As a Mac guy, I don’t use a great number of Windows software packages. However, the ones that I do use cover a pretty wide range of tasks, and based on my testing I’m confident of FLP’s software compatibility.
No compatibility problems with common software packages & full support for VirtualBox Guest Additions.
Here is the most important test for the viability of FLP as a remote virtual platform. While reductions in disk footprint and apparently wide software compatibility work in its favour, if it can’t deliver any noticeable reductions in CPU usage and speed then it won’t necessarily be worth the switch from XP Professional.
At idle, both XP Pro and FLP consumer around 12% of the host system’s RAM memory. The host server has 5GB of RAM, so a return of just over 10% is to be expected. After Windows has finished loading, FLP settles at 3% CPU usage. 3% appears to be the base line for XP Pro as well, however while FLP is stable at that value, XP occasionally leaps past 10%, presumably a result of the extra services running which are chopped from FLP.
Running Firefox, XP uses around 15% CPU at idle, however when browsing and scrolling this value increases to around 60-70%. FLP was a little lighter at idle with around 11%, but rose as high as 100% when loading multiple pages.
Of course, web browsing is not a particularly useful test for a remote VM, as it’s something you can do on any client OS. Having a Windows VM is most useful when you receive something like a Microsoft Visio document, for example. The VM provides compatibility with a filetype I can’t natively open and edit on my main machine – so let’s compare Visio performance under XP Pro and FLP. Here, XP Pro idles around 6%, and moves up to around 45-50% when editing a document. In this test, FLP was around 10% less resource-hungry under load, a useful difference if you were using multiple VMs on the same server.
Finally, the bootup time of the systems. With fewer services, FLP boots to its logon screen in 18 seconds from the execution of the VBoxHeadless command, in contrast with XP Pro’s 23 seconds.
Comparing one version of Windows XP to another is a bit of a strange exercise, however in the world of virtual machines and remote access, saving CPU cycles can make a noticeable difference in responsiveness and thus comparison is justified.
While Windows Fundamentals for Legacy PCs does save CPU resources (in most cases, anyway), the savings aren’t necessarily compelling enough to make an XP Pro user jealous. Of course, these savings in host resources, at least in my testing, were at zero cost to compatibility, and indeed came hand in hand with an enhanced installer, faster boot time and a smaller install footprint.
In that respect, FLP is a useful environment for virtual machines, and demonstrably superior to XP for this application – but only in situations where disk space and CPU power are limited.