As reported by The Industry Standard, eBay is buying, a large German on-line auto classifieds site. In case you didn't already know, has been a long-time user of MySQL on their site.

Not only do they run MySQL, but from what I've heard they've managed to push it very, very hard. Good stuff.

Posted by jzawodn at February 03, 2004 12:48 PM

Reader Comments
# sam said:

i love mysql, but the headline is a joke, right? a german classified site spurs ebay to move from oracle to mysql???

on February 4, 2004 08:26 AM
# Peter Pentchev said:

While I do recognize the above comment for the flamebait it is, let me just point out that Jeremy did not say that eBay is *moving* to MySQL. All he implied is that part of eBay's business will rely on MySQL for its operation, since (without knowing the details, so not really authoritative in any way) IMHO it would be a very, very unwise business decision on eBay's part to force to reimplement its web applications for no better reason that using a different database server.

on February 4, 2004 08:33 AM
# Dirk said:

If has done it right, it would be no problem to switch DBs.

If you stop accepting new cars/users whatever for an hour, this should be feasible.

on February 4, 2004 10:10 AM
# Mike Hillyer said:

But the question is why? If it aint broke don't fix it! And if it was broke, eBay wouldn't buy it in the first place.

on February 4, 2004 01:02 PM
# Justin said:

Mike, The phrase “If it ant broke then don’t fix it” does not really work in an IT world for most things. They might not have the technical expertise to troubleshoot / support the mysql environment once they make the switch.

You are now supporting two environments / two sets of hardware. If they moved the database to the existing infrastructure then they only support one system and one set of hardware.

It makes sense to stick with what you know and utilize the expensive hardware that you purchase.

on February 4, 2004 01:34 PM
# Mike Hillyer said:

Yes, but your argument only flies if we assume eBay fires everyone at Even if we assume a mass firing, you also have to factor the costs of conversion against having one of your employees learn MySQL and keeping the system that is already working.

on February 4, 2004 01:54 PM
# Justin said:

Mike, Typically they are not going to keep every one in the tech department to simple manage the additional services. You see if they merge the system then it is a little more overhead for their current staff to manage. They might not fire everyone or even more the data center but it would make financial sense to merge the system if they do more the service in-house to ebay.

If you keep both systems running then you have duplicate hardware, different problem, and twice the personnel. If they merge the systems they have only one hardware set, one team to manage the service, and possible only slightly more personnel resources.

It would seem to me that it makes more sense to merge the system and bite the extra cost of conversion than maintaining all the hardware / software / people to manage two systems.

But then again i'm no money man, not do i know if they will even move the hardware under one roof :)

on February 4, 2004 03:04 PM
# Justin said:

sorry jeremy - completely off topic, but there's this weird "thing" in the most recent Opportunity panorama image.

i checked out the original image on the MER site , and yup - it's really there.

Top image - hi-res is the best:

on February 4, 2004 04:04 PM
# Jeremy C. Wright said:

I'm sorry, but "if they write the code properly they could change" is just piss poor. It's proper resource management to optimize for the database you're using, thus getting the most out of it.

Even if you write a truly precious DAL, you're still better off optimizing for the DB you've already invested big resources in.

Also, believe it or not, the "if it ain't broke don't fix it" does fly in IT. Generally because people in IT don't make the real decisions, which is a good thing if you think that somehow it makes sense to switch to another DB for the sake of it (already have the staff and hardware, so it's not an acquirable cost).

Either way, legacy systems are a reality in any organization, something you learn to live with. Hell, we're still running systems from 1973 here at work... Because they aren't broke and there's no business case to update them.

on February 4, 2004 05:18 PM
# Justin said:

Jeremy, I think it is just poor planning to keep two systems live if you can merge them. IT might have some "legacy systems" simply because there was no need to change.. but if you are looking at saving more money (this seems to be the latest IT trend) then I bet it would happen.

Sure why not use the existing hardware / whatever when you are the only people using the stuff. but when you have a takeover and they already have existing hardware that can be running this website / database then why not? Really I think you are missing the point.

You bite the cost upfront of merging the systems and you save all the money of hardware / software / people resources. Sure it might not pay off in the beginning but think about the salary / upgrade costs / exta hidden costs over not just today but 6, 12, 24 months. Those could really add up to something. I think we might both be right about “if it ain’t broke don’t fix it”. Make sense? Then again i guess this is why allot of IT people find themselves out of work with a big wtf just happened.. Thinking the company is going to die without them.. when really is makes more sense to purge extra fat and save money in the long run.

but what do i know :(

on February 5, 2004 08:56 AM
# Mike Hillyer said:

Justin: When it comes right down to it you are neither right nor wrong, and Jeremy and I are in the same state. If you and I and Jeremy were all in the eBay IT department we could be sitting around a boardroom table tossing this back and forth and in the end neither of us could make an argument so convincing that the sheer merits of our arguments would get the decision made.

Lets Look at the issues: is currently handling traffic just fine as-is. Lets assume there are no security, scaling, or other issues causing us to have no choice but to migrate to Oracle.

If we were to migrate, we would not just be looking at the backend. I am sure that the actual server-side language eBay uses is different from the one uses.

Migration involves: New backend, new language, physical movement of servers from Germany ( and are in the same subnet so therefore is in the bay area, I doubt is.

Money saved: not supporting two architectures, less staff needed.

Money spent: time to migrate application, maybe some new hardware, more Oracle licenses, etc. Possible new staff in eBay campus to maintain

See what I just did up there? I made a lot of asumptions, which is the problem here: we do not know any of the facts, and as such this is all academic anyway. I will make one point: my new job is for a company that owns a bunch of ski resorts, most of them bought from others. Was our first move to go in and replace the systems they currently use with our own software? No. Why? Because with our limited resources it was easier to learn the other systems and integrate them rather than rip them out and replace them with our own. This is true in a lot of cases.

I like MySQL a lot, but If I inherited a large active site running Oracle I would not make my first act to fire the Oracle DBA and migrate the site to MySQL. Why? Because even if every other site in my posession is MySQL based, it is probably cheaper to keep the Oracle DBA and let him run that site than change it to PHP/MySQL. Opportunity costs are everywhere and have to be considered.

In the end it is up to eBay and all we can do is watch and wait.

on February 5, 2004 10:11 AM
# Jeremy C. Wright said:

The fact of the matter is that integration is becoming easier and easier.

I mean, in our environment we're tying ActiveDirectory / SMS in with eDirectory / ZENworks. That's a huge tie in by anyone's estimation.

Well, we're about to tie that setup in with OpenVMS (Alpha). Another huge tie in. I'm going to be in charge of the NT / Novell to VMS integration and I'm guessing it's going to take less than 40 hours of total work. Integrating something that complex would have taken thousands of hours just a few years ago.

Now it's pittance. And with no up front costs, the justification to do a switch becomes smaller and smaller.

Eventually we'll replace VMS. Not anytime soon though, as a huge amount of the hospital equipment runs on it. And unless you know of MRI machines that run Windows, I'm not sure of a way to speed up the process.

Mike's right. It's a per-case thing. Run the numbers. It's doubtful that a simple "replace it, stupid, because it's simpler" argument is likely to suffice.

on February 5, 2004 12:34 PM
# Jeremy C. Wright said:

It might sound like I'm anti-standardization, but really I'm very much for standardization. I just think it needs to be timed with business needs, otherwise you're creating un-needed expenses.

I mean, a simple example.

An organization has 2000 desktops. About 500 are 9x machines. Now, do you upgrade those machines to the standard (new hardware and software)? Even assuming you've got decent imaging going on (and standard hardware) so that you can just PXE-drop an image on there (no configuration), doing those 500 machines is still an incredible amount of work.

Does it make sense to make a project out of it, or to replace it "as you go along"?

I'm just not convinced standardization as an end unto itself is worthwhile, the numbers simply don't work. If it costs 55K to do the 500 Windows 9x machines now, but only costs 5K to do it "as you go along", you're likely to get your ROI faster, especially if "as you go along" is likely to happen in 6-12 months (and the "do it now" would take 3-6 anyways).

But, if someone can show me that doing those 500 machines is worthwhile... Well, we haven't installed the ZEN for Desktops client on 2000 of our desktops, so if you can make a case for 500 I might just get an award for making a case for 2000.

on February 5, 2004 02:38 PM
# Justin said:

I guess you need to weight in if the 9x machines are costing you more money to maintain?

If the 500 machines are crashing more / don't support the latest virus scanner and generally are slow and don't run the required apps.

Then you have to factor in what it would cost to upgrade / man hours / hardware upgrades or system replacements.. etc. desktops are hard because they don't really relate to the server side.

I guess the best example you could have.. is say you have 50 developers who have dual workstations.. win2k / win9x.. they use their win2k systems for everything.. e-mail, development, task management, etc.. they use their win9x box for some legacy system that their company once used.. maybe it is a black box type setup...

Now is it worth porting the win9x code to run on win2k so they can remove the win9x systems? Cut the admin costs / resources / power / desk space, etc... who makes the call and who provides who with the information that this is even possible?

I really have no idea how ebay is setup so i am really only guessing. I just wanted to be objective. I guess you have to factor in not just the IT side but also what the CFO thinks. Does IT tell the CFO what is happening or do they simply say nope... it is going to take more resources and ignore the factor that there is an option to merge the systems.

I guess this would have to go to some type of project manager to plan this all out and see what is the best option for their setup.. Really I have no idea what is going on!

on February 5, 2004 03:23 PM
# Justin said:

who cares about what ebay use. there's more important things going on - like the fact that the Easter Bunny is alive and well, and living on Mars. Now THATS a bit more important.

on February 5, 2004 03:36 PM
# Kevin said:

Good conversation you guys. Pretty intelligent ideas and arguements ! Nice !

on February 5, 2004 06:33 PM
# Mike Hillyer said:

"the Easter Bunny is alive and well, and living on Mars"

Now that is just silly! Everyone knows that the whole reason the Spirit rover was broken was because it ran over the Easter Bunny and got it stuck in the drivetrain... alive and well indeed!

on February 5, 2004 10:49 PM
# Dominik said:


first of all is a great side for selling cars, but the business volume is only 22.7 millions p.a. (source:
Ebay paid für 121 million Euro. If ebay doesn't change anything, it will take decades to make even.
So they will for sure, run and integrate the content in their own system.
for this they have to change the db-structure of other solutions doesn't make sense.

btw. sorry for my worse english

greetings from Germany

on February 6, 2004 03:09 AM
# Jeremy C. Wright said:

That doesn't make any sense. If they're looking for a short-term ROI, standardization isn't the way to do it. Increase visibility, drop the cost per unit that kind of thing. Standardization is never an end unto itself.

on February 6, 2004 05:16 AM
# Dominik said:

The reason for standardization usually is cost saving. You do not save any money, when you drive with two different systems. Ebay will use synergies in technic and labour and will try to maximise profit and market share.

And with a purchase price about 121 mil Euro nobody is looking for a short term ROI.


ps. I realy like ur Weblog. High level of quality.

on February 6, 2004 06:24 AM
# Jeremy C. Wright said:

Standardization's great when you're already doing something else along with it. Standardizing for it's own sake won't save you any money, it's just a make work project.

Especially when you're talking about databases when there are so many transformation services and applications out there that are much cheaper than simply standardizing.

In any organization with disparate applications or infrastructure there comes a time when standardizing makes sense. Always.

But to say it always makes sense is just too simplistic for me.

on February 6, 2004 09:25 AM
# jwr said:

There are too many Justins.

on February 7, 2004 12:03 AM
# Robin Schuil said:

Not only is using MySQL, (also an eBay company) is using it too. Even Kijiji uses some MySQL servers for their forum and calendar.
Marktplaats is running MySQL since 1999, now serving several hunderds of thousands users a day.

on October 14, 2005 02:44 AM
# Toby said:

Robin, I'm impressed with MySQL but I doubt it's enough to convince people yet.

on November 7, 2005 03:35 AM
# Auto verkaufen said:

Now at this time, the most use mysql.

on February 9, 2007 06:27 PM
# Alikzir said:

im a student...and want to know what was the cost of implementation paid by eBay to oracle when they moved to oracle 8i and then from 8i to 10g ?

on April 24, 2010 05:08 AM
# Alikzir said:

or lets say i want to know the price eBay paid to oracle for 8i and 10g
i need it for my presentation...

on April 24, 2010 05:21 AM
Disclaimer: The opinions expressed here are mine and mine alone. My current, past, or previous employers are not responsible for what I write here, the comments left by others, or the photos I may share. If you have questions, please contact me. Also, I am not a journalist or reporter. Don't "pitch" me.


Privacy: I do not share or publish the email addresses or IP addresses of anyone posting a comment here without consent. However, I do reserve the right to remove comments that are spammy, off-topic, or otherwise unsuitable based on my comment policy. In a few cases, I may leave spammy comments but remove any URLs they contain.