So PC-BSD and FreeBSD are getting a lot more attention. They are growing fast. Reports created based on the data from DistroWatch show that PC-BSD and FreeBSD are two of the fastest growing operating systems last year. Of course, despite my bias towards FreeBSD, as it is my favorite distribution, the numbers clearly show that Ubuntu/Debian-based platforms lead the growth.
Ok, so DistroWatch.com counts the hits per day (HPD) to a distro’s home page. Lets compare the hits per day over the past twelve months to the hits per day over the past 6, 3, 1 month intervals to see who is experience the most growth in hits per day as well as who has the highest percentage growth.
Growth in hits per day (HPD) between the 12 month and 1 month charts
Is one month a valid sample size? Of course not, that is why we are doing three month and six months as well. But lets look at it anyway.
PC-OS appears to have the lead here. PC-OS is based on Ubuntu (which itself is Debian-based). Other Ubuntu/Debian-based platforms showing growth are Debian itself, MEPIS, Mint, and Ultimate.
The only other base platform to have more than one distro show up in this list is FreeBSD. As you can see, PC-BSD is second in growth on both HPD and percentage, and FreeBSD is fifth in HPD growth and third in percent growth.
Growth in hits per day (HPD) between the 12 month and 3 month charts
Three months is definitely a larger sample size. Three months means we don’t have as much data skewed by release cycles which cause higher growth temporarily that will be offset by a decline the long months between release cycles.
Again we see similar trends in the three month reports.
Debian itself , MEPIS, Mint, and Ultimate and are all Debian/Ubuntu-based distros.
Again, FreeBSD and PC-BSD are on both the list. PC-BSD is second in HPD growth and leads all distributions in percentage growth.
Growth in hits per day (HPD) between the 12 month and 6 month charts
Ok, so the twelve to six month comparison is the largest sample size can get. Perhaps I should contact DistroWatch.com and see if I can get the raw data for multiple years past, but alas, I only pull the data from the tables they have currently available.
So now we see very similar data again. Seeing similar data a third time using this largest sample size means it is more likely accurate.
Fedora ties Ultimate for first, but the tie breaker has to go to Ubuntu/Debian-based plaftorms as they again lead Mint, Ubuntu, and Ultimate all on the list.
FreeBSD is on both HPD and percentage reports, while PC-BSD only shows up on the percentage report. However, again, FreeBSD is the only base open source operating system after Ubuntu/Debian-based to have two distributions show up between these two lists. For those interested, PC-BSD was fifteenth but only the top eight distros are displayed.
Copyright ® Rhyous.com – Linking to this article is allowed without permission and as many as ten lines of this article can be used along with this link. Any other use of this article is allowed only by permission of Rhyous.com.
Ok, so I am into FreeBSD and open source software, but I have recently had to do a QlikView implementation for my company LANDesk. QlikView has a feature where you can pull data from the web and report on it. So I needed to learn how to use the web-based reporting, so I decided to do a report from www.distrowatch.com.
Report Goals
There are few things that interests me from the data at DistroWatch:
Which base platforms are the most used?
Which platforms should software companies focus on supporting?
Where does BSD sit in the rankings.
How the report was made
So on the main DistroWatch page, there is a report that will give you the average hits per day (hpd) that a Distro’s web site gets. At the bottom there is a link to full popularity page of just these reports: http://distrowatch.com/stats.php?section=popularity
So at first glance, you see Ubuntu is the best and Fedora is second and so on. I wanted to take the statistics a bit further. I wanted to know what main base distribution was the most used. What I mean by base distro is this: Ubuntu is #1. But Ubuntu is not a base distribution, instead it is based on Debian. Mint is #3 and is also based on Debian. Debian itself is #6 and it is a base distribution. Fedora is a base distribution.
QlikView can connect to this web page and consume this data. It was also able to loop through and click go to the link for each distribution where it was able to pull the “Based on” results. I did a few little tweaks to clean it up.
So I used QlikView to match each Distribution to its base distribution and built my report. I gathered the cumulative hits per day (hpd) of each base distro by summing the hpd from itself and its child distros. The results are staggering.
Result of the Report
I am going to show you a screen shot of the report, but I am only going to show the top 10 base distributions because otherwise it is to hard to view the report.
# 1 – Debian
Well, I have to say that I new that Debian (13818 hpd) was popular because of Ubuntu, but I didn’t know how far ahead it was compared to other base distributions. I expected Red Hat to be a lot closer but its just not. Lets look at the top ten Debian platforms by hits. In QlikView this is easy, I can simply click on the Debian pillar in the report.
So not only is Debian’s cumulative hits per day first, but it is first by a long ways. The cumulative hits per day of distros based on Debian is more than three times larger than any other base distribution’s cumulative hits. It is pulling away from the pack and nobody is going to catch up any time soon.
What I don’t know is are these new users or are other distributions losing members to Debian or Debian-based distros?
You might be grumbling to yourself and saying some incorrect statement like: Well, Ubuntu doesn’t have Enterprise support like Red Hat. But like I said, that is an incorrect statement. See their support page: http://www.ubuntu.com/support
# 2 – Red Hat
Now, lets look at the top ten distros under Red Hat.
Ok, can I tell you that I was surprised at these results. I realize that Fedora was huge, I mean it is second on the distro list under Ubuntu, but I had missed the fact that CentOS was getting more than twice the hits Red Hat itself gets. The rest are hardly worth mentioning.
Historically mong Enterprise environments Red Hat is the most known distro, but when you look at these stats, you have to wonder if Ubuntu has taken over. The numbers for Fedora are fine, but for Red Hat they are not really that good. In fact, I keep hearing about companies using CentOS instead of Red Hat and as you can see, CentOS is getting a lot more exposure than Red Hat.
I will make this statement. Based on this data, if you are a software company considering whether to support Debian or Red Hat first, based on this data you have to choose Debian. If you were to make up some fuzzy logic for Red Hat (which due to its enterprise presence may or may not actually be valid) and weight the distributions based on other factors and somehow found a way to say Red Hat and its distro’s cumulative hits per day were worth three for every one, it would still be less than the cumulative hits per day Debian gets.
# 3 and #4 – Mandriva and Slackware
Ok, back to the report. Something that shocked me from the first chart and I had to analyze it further. Slackware? I had no idea that it was third. However, is it really third? It has a lot of very small distros based on it and Slackware itself gets 590 hpd and most the distros get less than 100 hpd. Mandriva is fourth but arguable could be third over Slackware. In fact, I have to call Mandriva third over Slackware. Sometimes you have to look at the data and make a judgment like this. Sorry Slackware, I am not trying to be biased (otherwise I would be talking up FreeBSD). I have no bias to any Linux distribution. I just say this based on the fact that Mandriva (1048 hpd) and the based-on-Mandriva version PCLinuxOS (773 hpd) both get more hits by a long way than Slackware’s top distros. The only reason Slackware got more hpd was because it has a lot of distros that were really small, while there were very few small distros based on Mandriva. The difference in the amount of small base distros is most likely due to the fact that Slackware is one of the oldest Linux distros, if not the oldest remaining distro, so naturally it has more distros based on it.
# 5 – Gentoo
Gentoo’s cumulative 1804 hpd was fourth. I have to apologize to Sabayon (760 hpd) as I had never heard of it until now. Gentoo itself only gets 428 hpd.
# 6 – BSD
What is next. Well, finally BSD shows up at number 6 with 1743 hpd. For those of you that are reading this and only know about Linux, BSD is NOT Linux. It does not run on the Linux kernel and is not likely to use many GNU tools. I hope I don’t drip with too much bias as FreeBSD is my favorite open source distribution.
Lets pull up the chart of BSD distros. There are 15 distributions listed under BSD, which is probably more than most people would believe since BSD often claims that it is not as broken up as Linux, but it has had its share of forks.
FreeBSD (553 hpd) is the main distribution. Of the Linux distributions, only Debian has more software packages available than FreeBSD.
PC-BSD (355 hpd) is to FreeBSD as Ubuntu is to Debian. For being such a new distribution PC-BSD is doing rather well. It is pretty comparable in ease of use to Ubuntu, Fedora, and OpenSUSE. Yes, PC-BSD is fully featured, running a nice windows environment with everything you could want, including a working Flash Player, the ability to configure your wireless card, and more. I recommend that if you are looking for a new desktop distribution, you at least install PC-BSD and give it a try. Ok, so my bias does show a little here.
# 7 – SUSE
So I was very surprised that SUSE wasn’t on this list until #7. Well, OpenSUSE is doing its part getting 1327 hpd. Remember, OpenSUSE is #4 if you just go by distro and not cumulative base distros. I think in time SUSE could be more popular. SUSE is newer than some of the other base distros and so it only has four distros listed. Novell’s SUSE Linux Enterprise (121 hpd) is the second most popular SUSE distro, however, it just not getting any were near the hits I expected it to be getting.
The others
And then there are the rest of the top ten: #8 Arch, #9 Puppy, and #10 Solaris (Or is that Oracle now?). Sorry if your distro was left out, this report is in the control of those who visit the distro’s web pages.
How accurate is this data?
On DistroWatch’s popularity page, it says:
The Page Hit Ranking statistics have attracted plenty of attention and feedback. Originally, each distribution-specific page was pure HTML with a third-party counter at the bottom to monitor interest of visitors. Later the pages were transformed into plain text files with PHP generating all the HTML code, but the original counter remained unchanged. In May 2004 the site switched from publicly viewable third-party counters to internal counters. This was prompted by a continuous abuse of the counters by a handful of undisciplined individuals who had confused DistroWatch with a voting station. The counters are no longer displayed on the individual distributions pages, but all visits (on the main site, as well as on mirrors) are logged. Only one hit per IP address per day is counted.
There are other factors to consider, such as the fact that some of the distributions are Live CD distros and not really platforms meant to be installed. It would be interesting to exclude them and only include installable distros but for lack of time, I didn’t.
I did nothing to verify the accuracy of the data at DistroWatch and any errors you see are not likely mine, as all the data was pulled from DistroWatch, please report any error to them and once they fix these errors, the QlikView report’s data can be reloaded.
Also, this data includes all hits from all areas: Consumer, Enterprise, Education, etc. Unfortunately there is no way I know of to tell where the hits came from. If there is a distribution that is 100% education hits, there would be no way to know that. Obviously if your target is Enterprise, you are left wondering which open source distros are really the most used in Enterprise environments. Unfortunately this report doesn’t answer that question. This is not a report of installed platforms, it is a report of cumulative hits per day. It is what it is.
Copyright ® Rhyous.com – Linking to this article is allowed without permission and as many as ten lines of this article can be used along with this link. Any other use of this article is allowed only by permission of Rhyous.com.
Ok, so the fact that I cannot just have one line in a Load Script is a negative for QlikView. In a perfect world, I would have one line that would give me a bunch of possible values I could use for a dimension, such as CalendarDay, CalendarWeek, CalendarMonth, CalendarQuarter, CalendarYear, etc… It would be one line like this:
Calendar(StartDate, EndDate);
Alas…it is not a perfect world, so this feature doesn’t exist in QlikView. (Enhancement Request please!!!!)
So there is a Wiki on how to do it. Here is the link.
http://community.qlikview.com/wikis/qlikview-wiki/how-to-create-a-calendar.aspx
However, the problem is that this didn’t work.
So after some research I remember that internet search engines exist and I don’t have to just search QlikView’s site and documentation. I did a google search for this string:
qlikview how to create a calendar
The script didn’t fail to load…yeah…wait…there is not data in my report that has to do with a Calendar.
This sucks. Why can’t I just create a new Calendar. This is common problem with some software companies. There is a “key features” that can be done, but with great difficulty. However, because it can be done, they don’t spend any more development time on it.
LET vDateMin = Num(MakeDate(2000,1,1));
LET vDateMax = Floor(YearEnd(AddMonths(Today(), 12)));
LET vDateToday = Num(Today());
TempCalendar:
LOAD
$(vDateMin) + RowNo() – 1 AS DateNumber,
Date($(vDateMin) + RowNo() – 1) AS TempDate
AUTOGENERATE 1
WHILE $(vDateMin)+IterNo()-1<= $(vDateMax);
MasterCalendar:
LOAD
TempDate AS CalendarDate,
Day(TempDate) AS CalendarDay,
WeekDay(TempDate) AS CalendarWeekDay,
Week(TempDate) AS CalendarWeek,
Month(TempDate) AS CalendarMonth,
Year(TempDate) AS CalendarYear,
'Q' & Ceil(Month(TempDate)/3) AS CalendarQuarter,
WeekDay(TempDate) & '-' & Year(TempDate) AS CalendarWeekAndYear,
Month(TempDate) & '-' & Year(TempDate) AS CalendarMonthAndYear
RESIDENT TempCalendar ORDER BY TempDate ASC;
DROP TABLE TempCalendar;
LET vDateMin = Num(MakeDate(2000,1,1));
LET vDateMax = Floor(YearEnd(AddMonths(Today(), 12)));
LET vDateToday = Num(Today());
[/sourcecode]
Now when your script loads, you can right click and choose New Sheet Object, Slider/Calendar Object.
Choose Calender, not Slider and base it off of the CalendarDate field. Also on the Sort tab, use the Numeric Value to change the sort to Descending.
