← Back to team overview

pbxt-discuss team mailing list archive

Re: State of PBXT

 

Indeed I won't be at the conference this year either. In part due to PBXT's absence, but also due to topic selection among other things. Oh well. Maybe next time!

On Mar 9, 2012, at 5:27 PM, Paul McCullagh wrote:

> In the end it always comes down to the business case. We can't be idealistic about that.
> 
> The bottom line on PBXT is that it wasn't a loss, but we didn't make much of a profit either.
> 
> But that was not the interesting part. The real interesting part was working together with MySQL and Sun, and we were doing that.
> 
> Just imagine if MySQL had IPO'ed. Just imagine if Oracle had not bought Sun...
> 
> We could all see that Falcon was struggling, so the potential was there. 
> 
> We knew about the potential on that side, and speculated.
> 
> But, if that is the bottom line, then I am pleased to say, there is a bonus on top of it all. I have met a lot of great people and made some very good friends!
> 
> I won't be at the Conference this year, but I know time is not a factor.
> 
> We'll, meet again and talk about the "old times" soon enough ;)
> 
> On Mar 10, 2012, at 12:04 AM, Tim Soderstrom wrote:
> 
>> 
>> On Mar 9, 2012, at 4:47 PM, MARK CALLAGHAN wrote:
>> 
>>>> 
>>>> Perhaps but I assume you're not running Facebook on DropBox :) Certainly there is still interesting use cases for RDBMS's and the engines therein. InnoDB has caught up so much that PBXT no longer has the edge, but I still find it very interesting. Were Paul to have the resources InnoDB does, I suspect PBXT would be significant.
>>> 
>>> I agree with you. If there were more resources behind PBXT it could
>>> have been very interesting. I don't know how much of it is
>>> log-structured compared to the original design but I was very
>>> interested in that. I blame myself and others who didn't have time to
>>> help Paul get this to market. I suspect that PBXT still has advantages
>>> compared to InnoDB on multi-core servers.
>> 
>> The log structure was very very cool, but I actually liked the hybrid approach Paul was doing with the fixed rows + log structure. It was quite elegant, although, being a casual observer in the whole process, it's hard for me to see the real impact of the design. Still, I found it very very cool.
> 



References