VMWare Fusion 3.0 and Parallels Desktop 5

Long ago, when VMWare Fusion was first announced, Parallels Desktop was the only way of running Windows on a Mac (I think this was even before Boot Camp came along). VMWare was the sort-of goliath virtualization company, having long since been making Windows and Linux based virtualization solutions. I remember downloading the trial of Parallels and feeling annoyed that I had to do little tweaks to make things work properly and that they lacked support through their online forum. I reached out to the VMWare Fusion beta, which was free until final release, and found stability, simplicity and best of all, their engineers actually responded on their forum! I bookmarked that very link and decided right then and there – Fusion was my choice for Windows on the Mac.

Things have changed a lot since then. Microsoft released 7, finally a respectable Windows operating system, meaning that the cautious folks can finally upgrade from Windows XP. Vista and 7 have some excellent DirectX-based user interface (UI) animations that make working with the UI a pleasant experience, but of course virtualizing these features was a long-time in the making. The Parallels folks started responding to people in their forums and their downloadable application just started working right after download – without requiring tinkering with the settings. The lines between the two applications began to blur, both in terms of features and support.

Despite this blurring of differences between the two products, when I heard the announcement for VMWare Fusion 3, I thought, finally! Aero will work and I’ll be able to upgrade for a good price ($39.99 rather than $79.99 for the full version). I ignored Parallels Desktop 5, thinking that I could get the same thing cheaper with my upgrade copy of Fusion. Was I wrong? I certainly didn’t think so at the time – I thought, how could VMWare have released 3.0 without it performing similarly (or better) compared to Parallels Desktop 5? I simply didn’t believe they would do that.

Here we are, a week later, and Parallels Desktop 5 is out. I’ve already paid for, downloaded and installed my copy of VMWare Fusion 3. I love the faster resume and suspend times and am enjoying the Aero effects in Windows 7, but I’ve noticed some problems. After all these months of working on Fusion 3, the performance (excluding the aforementioned suspend/resume times) feels the same as it was with Windows 7 on Fusion 2. This is by no means an actual benchmark, simply what working with Windows 7 running in Fusion 3 with 1GB of RAM and a single core feels like to me – the end user. Being of the belief that the VMWare engineers tweaked and tuned the hell out of Fusion 3.0, I took this to be the best they could do, and assumed that Parallels couldn’t have done better (or at least, if they did better, it would be negligible).

Then I read this post from list-whisper.com – admittedly an unknown source to me, however the writers words sound sincere and I have no reason to doubt them. And then came Ars Technica. Ars Technica is not just a well known source for news, but the most informative IT news source I’ve found on the web. I’ve been reading them for years and I put a fair amount of trust in their collective opinions and reviews. The Ars article describes Parallels Desktop 5 as being much faster than VMWare Fusion – the catch? They compared it to VMWare Fusion 2 – not a truly fair comparison.

Interestingly eWEEK.com indicate that their “testers preferred VMWare Fusion 3 to an earlier version of the Parallels solution.” But again we find mismatched comparisons between versions and they don’t discuss whether or not their preference comes from performance, a feature-for-feature comparison, stability, or whatever other methods they might have used to compare the two products.

Andrew Bednarz from theappleblog.com writes “More importantly than all of that, its super fast. For me, it’s much much faster than VMWare Fusion 3.” Add to this the number of tweets that @ParallelsMac has retweeted describing how much faster it is. Of course Parallels would only retweet the positive ones, but still, there have been more than enough tweets and reviews to give it credence. I have seen nothing of the sort from VMWare users, and my own experience matches the social media outcry perfectly.

Sure, I could ignore the reviews that directly compare the two and come out with Parallels Desktop on top, leaving only those describing Parallels 5 being faster over previous versions of Parallels. But even in doing that, we get users of Parallels 5 describing the major performance improvements over Parallels 4. I haven’t seen any such thing from users of Fusion 3 exclaiming their surprise with the newfound speed over Fusion 2 (I certainly didn’t). This leaves us with one question – was Parallels 4 slower than Fusion 2 to the extent that the perceived differences to the upgraders were shocking for Parallels users vs. Fusion users? Not according to the benchmarks [mactech.com].

For the time being, it does seem that I have chosen the wrong application – I should have switched to Parallels before I spent the money on VMWare Fusion 3. I believe that the stability problems associated with Parallels Desktop that the lost-whisper article describes can be solved rather easily, at least when compared to the amount of work required to optimize performance of VMWare Fusion for better performance. Because of this, I won’t be expecting much from VMWare Fusion 3 updates along the lines of performance, however I will definitely be hoping that they will come so that Fusion will once again provide stiff competition for Parallels Desktop in the Mac virtualization arena.

As an aside, I will mention that there are features other than performance that will likely benefit users of VMWare Fusion over Parallels, however in most of the reviews it is described that Parallels provides more features than does Fusion. This doesn’t mean that Parallels does the equivalent features better (ie: Coherenece vs. Unity), just that the general consensus is that Parallels Desktop 5 has more of them. In terms of Coherence vs. Unity, from the reviews I read, it appears that Coherence is actually doing a better job – with less ‘wiping’ effect (where the background from Windows shows through when you move windows around) and better corner rounding to make the windows feel more Mac-like, although I have not used Parallels desktop 5 myself to confirm that. I will update here when I have more!