Which configuration manatement tool is more popular: Puppet or Chef? RedMonk analyst Stephen O’Grady recently did some Web mining to find out. O’Grady’s not sure. I’m betting that Puppet has more traction, but that Chef is growing fast.

O’Grady explains the project thus:

Although discussions of the platforms’ relative technical merits can be interesting – the comments on this HN thread display the usual range of opinions on the subject – we’re typically more interested in usage patterns. Quality of implementation is an important consideration in technology selection, but history demonstrates adequately that technically inferior solutions can and often do outperform competitors. Because there is no single canonical source for usage, we instead examine a variety of proxy metrics, looking for patterns that indicate a broader narrative at work. Here’s a non-comprehensive run through some of the metrics that we regularly evaluate.

RedMonk’s Metrics

Unfortunately, none of these metrics gives a clear idea of how much traction either tool has. O’Grady looks at Debian installs, where Puppet has an enormous lead, but Opscode, the company behind Chef, notes that since most people install Chef as a Ruby gem, not through a repository, this data source gives no clear indication of Chef’s traction.

O’Grady also looks at Hacker News mentions, where he finds the two tools have a similar number of mentions, but points out that Chef has mentions starting well before the project was actually released. I’m not sure how O’Grady controlled for false positive results, but I’m guessing that “puppet” and “chef” might not be easy keywords to mine.

The data he found on Github is interesting, where Puppet leads in project views but Chef leads in engagements like forking, but not a good indicator of actual traction for either project.

As an afterthought he says that the number of mentions on StackOverflow are comparable, but this is hardly a good way to judge the tools’ traction. It could be that one tool has more users but fewer questions and that the other has more questions and fewer users.

The number of shared scripts is also of murky value. Opscode has more Chef scripts on its site, but there are more shared Puppet scripts on Github. It all evens out but doesn’t give a clear idea of the actualy usage of either tool.

Other Metrics

There have been a couple other comparisons of Puppet and Chef, one by Bryan Berry in October 2011. Berry compared the number of users in each projects’ respective IRC channel and mailing list. I’ve just checked these, and Puppet’s IRC channel had 673 logged in users and its mailing list had 3,860 members. Chef’s IRC channel had 324 and its mailing list had 776 members. When Berry looked Puppet had 541 users in its IRC channel and 3363 members on its mailing list. Chef had 233 IRC users and 600 mailing list members.

That would indicate that Puppet has a significantly larger community, and I think that’s a better, but not perfect, indicator of actual usage.

We can also use the mailing list numbers to estimate the growth of the communities (I won’t measure the IRC usage since I didn’t check at the same time of day as Berry did). Puppet grew by about 14.78% and Chef grew by about 29.33% since October – but of course Chef is starting from a smaller number.

Also, Dice lists 318 jobs mentioning the word “puppet” and 212 for “chef.” Since Dice lists primarily tech jobs I think it’s safe to say that these listing aren’t for puppeteers or cooks. That’s a pretty wide lead for Puppet, but considering that Chef is newer it’s doing quite well.

Does It Matter?

For those picking up job skills, it clearly does. Puppet seems to have the edge for now, and has great job growth on Indeed. But the Chef community is growing fast, Opscode boasts great corporate uptake and has a solid number of mentions on Dice. For now if you’re picking one or the other it makes the most sense to pick the best tool for the job as it seems that neither one is going anywhere.