Sunday 7 October 2012

No I'm NOT the only person in the UK using CF10!

G'day:
OK so that title seems like I'm stating the obvious, but it's a reaction to my earlier article which questioned whether I was. I was questioning this because Rakshith from Adobe had suggested I was, as far as he knew. I didn't believe him, so I decided to try to find out (by running a survey).

Firstly: thanks to everyone who responded.  I got 56 responses which is quite good for the surveys I run.

Secondly: yes, I can confirm that there are people in the UK other than myself using CF10.  Phew.


Here's the analysis of the data you gave me.  Just a note up front: I had a respondent from each of Russia and Mexico, which weren't on my country list.  Sorry about that.  And sorry even more than everyone not from either the UK or the States has been squished into "All" for each of questions below.  I did this because I was mostly interested in the UK, but about 75% of my readership is from the States, so I figured I'd break-out their figures too.  Also the individual country figures outside of the US and UK were mininal, so there didn't seem a point in treating them separately.  FYI, the nationality breakdown of the respondents was as follows:

CountryPercentageCount
United States39.2%22
United Kingdom37.5%21
Canada7.1%4
Australia5.4%3
Germany5.4%3
Mexico1.8%1
Netherlands1.8%1
Russia1.8%1



ColdFusion 10 server install base

United Kingdom




United States




Worldwide



This is a breakdown of the distribution of how people say they're using CF10.  There's a wee bit in production, but it's mostly still just people messing around with it.  I expected the proportions to indicate slightly more production use, but I guess all that stuff that's in pre-production will be going live before too long.

As always, there were comments:

We don't have any plans to upgrade to 10. We are moving to ASP.Net MVC.



We will eventually look at CF10 but not till next year as what we budgeted for it in 2012 did not take into account the enormous price hike with the new licensing.



I'd have some stuff in production if I could find a free CF10 host.



Have never used CF10 and given the expense I have no incentive.



the price is just becoming too much of a factor. We've paid for CF Enterprise 9, but will pursue Railo/OpenBD moving forward and then start migrating to another platform. Love CFML, but not sure why it's our job - or even in our interests - to be loyalists to what is basically just a tool... and a costly one at that. I think that's just bad business... it reminds me of how people get about politics: rather than holding either side's feet to the flames over an issue, we just cheer for one side over the other as if it were a sports team. Silly. Like politics, a business's tools ought to not be based on emotion or loyalty, IMHO that's just bad business. CF no longer does what others don't, nor does it do things any longer better/quicker/more efficiently than its competitors. It's not personal, just business.
The middle one is my own comment (ie: the "I'd have some stuff in production[...]" one).

The move to ASP.Net MVC is an interesting one.

And I wonder how many other people have looked at the new licensing model and gone "nuh-uh".  I can possibly understand the new model for new purchases, but it really sux that Adobe have applied it to their existing customers wanting to upgrade.  As I understand it, it's basically doubling the cost for most people.  And for what?  From my perspective the merits of CF10 over CF9 perhaps warranted paying a normal upgrade cost, but not one that means buying additional licensing because Adobe happened to decide to change the way CPUs were licensed.  This seems wrong.



Non-CF10 CFML server versions


This is for the people who aren't using CF10 yet, and asking what they were using instead. Well: some people answered yes to both, which was a bit odd, but if one takes the questions separately, those answers are still meaningful, so I included them.

United Kingdom




United States




Worldwide


No real surprises here: it's mostly CF9 with a bit of residual CF8. And some clunky old CFMX6+7 too.  Yuck.  At least no-one's using CF5 any more! It's interesting that no-one from the States was using Railo or OpenBD!  Perhaps this is because this is a ColdFusion-centric blog, so I don't get much interest from them.  I was kinda hoping this blog was fairly platform-neutral... or at least I spend a "proportionately-balanced" amount of time on each platform?  I dunno.

Railo seems to be doing "better" than OpenBD (in that there's more "bums on seats", it seems).  In my limited experience this matches up with the activity I see in each of the communities... the OpenBD Google Group can seem very quiet at times, and their Twitter traffic is basically non-existent.  TBH I myself don't do so much of my testing on OpenBD any more because whilst Railo is pretty up-to-date in regards to CF compat, so I don't have to horse-around too much (or at all) with my test code to get it to run on Railo too, I seemingly always have to mess with the code to get OpenBD to work.  It seems to be slipping behind the game a bit.  Which, in turn I think, makes it less of an attractive prospect to people looking for an alternative to CF.  This is a bit of a shame.  Although maybe it'll just make the Railo community stronger, which is a good thing.

There were a coupla same-themed comments against this question:

CF9 currently, but Railo is on the development server and we're hoping to switch soon.





CF9 on front-end servers, Railo on dedicated task servers.

So instead of looking towards sticking with ColdFusion and going to ColdFusion 10... Railo is seeming like an appealing option.


Production farm size

This one was to get a feel for the scale of what servers were in production.  I've spilt the results out into "non-CF10" (the first three charts) and "CF10" (the second three), as that seemed more useful for this survey.

United Kingdom





United States





Worldwide





CF10 Production farm size

United Kingdom





United States





Worldwide



Basically most farms are 1-10-ish servers, with the farms larger than that being outlier results.  That seems reasonable.

And as for CF10 usage... yay!... there's at least a coupla production servers in the UK running CF10. And obviously my readership is very small, and also comprising a specific sort of CF developer (ie: one's that read CF blogs), so I'm sure there's a few more out there too.

However it's not all good news, as is borne out by some of the comments:


Won't be going to cf10. Too expensive. Possibly Railo or change altogether. Looking at dot net mvc and Lift. Latter is interesting but hell of a learning curve and poor documentation.





CF 10 support in our CMS was slowing down our the roll out. Now it's just every other piece of work getting in the way.





Railo hosted on AWS / GAE.





We are on a shared server.





We've decided to skip CF10 altogether and with Adobe's list of screw-ups we're even toying with the idea of abandoning it in favor of Railo 4 altogether.





We're waiting for 10.1





of that amount only 3 servers are under cf8 not even cf9... Client didn't see the reason to invest more on it.





no desire to upgrade them to cf10

If I was running a corporate site, I'd definitely be in the "waiting for 10.1" bracket.  This is partly a comment on Adobe's QA, but most down to being a comment on the IT industry's QA.  It's a bit rubbish these days.  And it seems a lot of a software product's testing is done on live users, and this is seen as acceptable.  It's not acceptable to me, so I'd rather hang around and wait until the first round of bugs are ironed-out by the early adopters.

I think this is a particularly sensible approach with CF10 with it's underlying-architecture change from JRun to Tomcat, and also applying the hindsight that the number of people with performance problems running CF10 on IIS  that I've heard about.  CF10 needs ironing-out.

For my own purposes my stuff is very low scale, so it it wouldn't really matter about performance, but also I use a shared host, so if they were to think about CF10, they'd have to be confident that it actually works properly under load, which seems questionable at the moment.

So that's the end of that survey... I've had one more response whilst writing this, and I've included their answers in here, but I'll take it down now.

I think the most significant things to take from this are:
  • people seem content with CF9, and possibly don't see a reason to upgrade. Are REST web services, and closures and that sort of thing not so interesting in "the real world"?  What sort of things would make people want to upgrade?
  • There is a bit of a backlash about the licensing changes in CF10, meaning it costs a prohibitive amount to upgrade. Adobe should look at this.
  • Railo is looking like a competitive upgrade alternative to CF10.  I suspect this is part pricing, and part the fact that Railo just seem really responsive compared to Adobe, and this is important to people.  For my part: they fix the bugs I raise.  And I don't even use Railo (and they know this)... they just have pride in their product and think it should work.  I don't get this feeling from Adobe, which is actually a consideration.
  • But people are using CF10 in the UK.  Phew ;-)
I'm running out of Sunday... I seem to have been sitting in front of this blimin' laptop all day, with one thing or another.  Time to do something else, I think.

Righto.

--
Adam