Is MySQL better than PostgreSQL? A debate on Enterprise Readiness
Recently we were approached by some of our users questioning our assessment/rating of MySQL versus PostgreSQL. Currently MySQL is rated with four stars for Enterprise Readiness, while PostgreSQL only shows three. All the other ratings (functionality, maturity, community, trend) are the same otherwise. This seems to be a good occasion to talk a bit about our rating criteria and the way we apply them. Also, we were having quite a bit of a debate in the newly formed EOS Advisory and Expert Board on this same topic, so it’s good to share some of the findings.
So, here’s how we describe the Enterprise Readiness rating on this site:
On the basis of the other criteria and additional experiences as well as further product characteristics (e.g. how easily a base technology be introduced can into the typical enterprise, how reactive the community is, how easily a product can be integrated in commonly-found enterprise IT environments, or how well does the product support open standards) the Enterprise Readiness Rating (aka “Optaros rating”) indicator is consolidated. This rating describes how capable an open source product is to cope with the needs and requirements of midsize and large enterprises and organizations. The EOS Directory does not list products that do not at least meet the 1-star rating.
To go a bit more in detail there are a number of aspects that influence our assessment of Enterprise Readiness, beyond of what is coming from the other rating factors:
- Popularity, market reach, distribution power, public awareness, typical ranking in top x lists:
Clearly here MySQL seems to outperform PostgreSQL. If you google for example for the two technologies you will see that you have approx. 10 times more hits for MySQL than for PostgreSQL. Same effect when you look at Google Trend.
Asking a typical enterprise architect what open source databases come to his mind, he usually will mention MySQL first. Even on EOS Directory MySQL is much more popular (number of page hits) than PostgreSQL.
- Availability and quality of professional services and consulting:
How easiy is it to find a training for MySQL versus PostgreSQL? How many potential partners can an Enterprise find to help with tuning, integration, support, etc.? Also here MySQL seems to be ahead of PostgreSQL. But with EnterpriseDB, Fujitsu, CommandPrompt, 2ndQuadrant, Cybertec and other firms, Enterprises should be able to find good help for PostgreSQL also. So, no real advantages of one against the other. - Adoption in the market (byEnterprises as well as SW vendors):
MySQL has been very successful in partnering with all kinds of SW vendors (not only Open Source product companies). This results into a much higher adoption of MySQL in both the software vendors and consequently their end customer. Even on SourceForge MySQL is referenced by more than 7′000 other projects, while PostgreSQL is being mentioned by less than 1′000. Independently many Enterprises have established MySQL clearly as their second database standard (after Oracle or IBM), this is less often the case with PostgreSQL. So in terms of adoption MySQL takes the lead again. - Enterprise specific atribute, features, extensions and requirements:
Here we look whether MySQL offers better/more migration tools for example, whether it’s easier to manage with the typical enterprise tools already in place, etc. MySQL may have a bit of an advantage here, but not by much. - Enterprise culture orientation:
Is MySQL easier to buy for Enterprises than PostgreSQL? Does the support of Sun (and now Oracle) help to make it easier to evaluate and consume? Probably yes, but with the disadvantage of being less “open source”.
Now there’s a lot more to say about the two technologies of course. There are some known problems with MySQL that ask for workarounds, similar issues probably exist around PostgreSQL. There are open source projects that clearly recommend PostgreSQL over MySQL, e.g. Jackrabbit or Django. These things however should rather influence the “maturity” rating than the Enterprise Readiness. So if we keep the top rating for Maturity for both technologies we can’t make this a differencing factor in Enterprise Readiness. The same is true with other aspects such as the validity of the community or the availability of features.
There have been discussions in the EOS Advisory and Expert Board to automate more of the rating process and base it on available data. This may be well the way to go for the future, but in my eyes it’s exactly these discussions and real world experiences that make the EOS Enterprise Readiness rating so powerful.
Now, to come back to MySQL and PostgreSQL, should we downgrade MySQL to three stars, upgrade PostgreSQL to four stars or leave it as it is today? Stay tuned, we will make our decisions in the coming days 
And anyway, both are good and widely used technologies. The difference isn’t big, but may be somewhat prestigious. Join the discussion!
As part of the news bulletin below we have additional information that is being added that may be beneficial to your open source community business. Effective 2026 we are introducing new best rated service provider solutions for cbd, cannabis, and smoke shops.
This entry was posted by Bruno von Rotz on Thursday, June 25th, 2009 at 5:38 am and is filed under Community, EOS candidates, Enterprise open source, Making of EOS, Open Source projects, User feedback. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.
Entries (RSS)
June 26th, 2009 at 4:00 am
Points 1) and 3) are closely correlated, and nobody would (currently) question MySQL’s leadership there.
As to point 4), I’d say that your assessment is debatable, particularly when features providing “enterprise readiness” are expected to exhibit certain standards of reliability and stability.
Concerning 5), I get the impression that something that cannot be bought commercially will be rated worse there. Ist that reasonable? Has Oracle shown flag how deep their commitment to MySQL runs?
June 27th, 2009 at 12:19 pm
I think Enterprise Readiness as a reflection of popularity for open source applications is valid, however the question goes beyond just what is the most popular technology for a given field, but rather it’s a question of reaching a plateau where the software becomes essentially equal to the individual consumer.
If you compare PHP and Python (both 4 star rated in your guide) in google trends, the numbers correlate with the MySQL/Postgres relationship, with PHP being much more popular than Python. However both languages have a large user base, have many supporting companies involved, have gone through several major releases, are in use by large enterprise shops, and have large established presences within the industry. These are all true of MySQL and Postgres as well; they have all reached the plateau, where any given enterprise should be able to find adequate support whatever there needs are. As a contrast, you can look at something like Erlang or couchdb. They have both been used by a few notable companies and are gaining momentum, but they have not reached the level of the aforementioned software.
So I think the focus shouldn’t be on Postgres vs. MySQL, a trap people have fallen into for years. The question is, has Postgres reached a level where questions about support companies, developer mindshare, project stability, and long term software maintenance have well defined answers.
By the way, I would encourage people to read up on the “Open Source Maturity Model”, which is a set of criteria that can be used to judge the enterprise readiness of software solutions. If you’re considering deploying open source software in your enterprise (or in the business of evaluating software for others), it lays out many of the criteria you should be using in your evaluations.
June 27th, 2009 at 7:17 pm
Thanks for explaining your ratings!
I do think that the board should have more discussion about this internally. I can see a couple issues with your rating system which limits its value to IT buyers:
1) The current rating is 40% based on pure popularity. While breadth of adoption is obviously important to enterprises in terms of being able to recruit talent and find training and tools, having 2 out of 5 starts be entirely dependant on popularity seems disproportionate. More importantly for EOS’s viability as a web site, telling people about the most popular projects doesn’t tell anyone something they don’t already know. For EOS to be valuable to an IT manager, they want to know about the enterprise-ready projects they *didn’t* already hear about.
In other words, if EOS is just going to tell me what’s the most popular, why do I need EOS? I can get that from Google and Freshmeat.
2) Your rating of Enterprise-Readiness does *not* include any application-specific measures of readiness. For databases, for example, you’re not checking stability, reliability, availability of bugfixes, scalability, enterprise features, BI or ability to match features provided by the industry leaders. In other words, when you rate a database for enterpriseness, you’re not really bothering to check database features, at all. This makes EOS vastly inferior to Forrester in terms of the value of your ratings — Forrester *does* research all of these things.
Even in the OpenBRR project, we have ratings for the specific application space and its internal metrics.
I think that both of the above need to be addessed if the “EOSDirectory” is actually going to have any utility to enterprise IT buyers.
June 28th, 2009 at 1:36 am
I do appreciate this discussion very much. Let me add a few comments to what has been said.
First, the “Enterprise Readiness Rating” is not based on an arithmetical formula adding up the subcriteria (e.g. popularity) listed. Subcriteria are seen as indicators, the overall Enterprise Readiness rating is created by looking at all the different points and then deciding whether we think the project overall is good enough to be rated with 1, 2,3 or 4 stars.
Secondly, besides the specific Enterprise Readiness subcriteria also the four other ratings (functionality, maturity, community, trend) are taken into account. So, for example, a project that can’t compete at all with the existing commercial proprietary alternative in terms of functionality, will not receive 4 stars on Enterprise Readiness neither.
Thirdly, when talking about the rating then it’s about assessing the project as such (i.e. looking at the rating defiinitions given under )/about/criteria) but also about comparing the project and its rating against the alternatives listed on EOS Directory. This is, as said before, not a mathematical process, but rather the result of discussions and debates.