Software patents is dangerous territory for developers. Typically advice I've been given is "ignorance is the best defense" - when you are in a patent fight, claiming ignorance of the prior art is a defensible position. Digging deeply into the mapreduce patent and fully investigating the limits of their prior art is an exercise I will leave to patent lawyers and other people who are not actively engaged in development. And that is my stand, and why I am indifferent.
Besides which, most software patents tend to have a fairly narrow scope, and small changes can avoid the wrath. -ryan On Fri, Jan 22, 2010 at 12:29 AM, Laurence Hubert <laurence.hub...@free.fr> wrote: > Hi Andry, Bruce, > > > >> It's a fair argument that Hadoop mapreduce is a Google MapReduce clone -- >> which >> has been extended in several directions by the community, of course. Given >> that, >> and the in my personal experience substantial commercial application of >> the >> technology already, clearly the success story has already spread far >> beyond any >> reasonable definition of "academic". > > Absolutely true. I am the proof of this :-) It turns out that I am a BIG FAN > of both > Hadoop and Hbase and I have a true willingness to use them in a real > business ... and > as soon as this business develops commit development resources to it. So I > am not > questioning the value of the work because I know this is great work and I > know the > adoption is going to be huge. > > >> I am also of the opinion that this is a defensive move by Google. It would >> be in >> direct opposition to years of experience we have with this company should >> they >> go after the ASF in any way. > > I tend to agree (and with Bruce as well). Still from a business perspective > this is a risk > and I hate risks :-( > >> Furthermore, I'm not a patent lawyer, but I believe for the HBase case, >> that >> HBase (and HDFS) are not covered by these patents, except for the >> mapreduce >> integration package, which could be dropped without any loss of HBase >> functionality for clients using the HBase client API. So the direct impact >> on >> HBase for some worst-case scenario would be low as far as I can see. > > Unfortunately when you use HBase this is because you have to process large > amount of data > which means that by nature many of the HBase adopters are ALSO hadoop > adopters. > But thank you for your comment because somehow it helps understand the > degree at which > we would be impacted if Google wanted to enforce its intellectual property. > > Thanks Andy and Bruce, somehow the discussion helped. > > Laurence > > ----- Original Message ---- >> >> From: Laurence Hubert <laurence.hub...@free.fr> >> To: hbase-dev@hadoop.apache.org >> Sent: Thu, January 21, 2010 7:33:29 PM >> Subject: Re: Google patent over Map Reduce - Hbase reflections >> >> Dear all, >> >> I think the impact of this patent should not be underestimated. If >> Hadoop/HBase >> is only an educational system and not used by anybody for any business >> then I >> agree there is no threat to the community... >> but if companies are relaying on it to do business (and some started to >> evaluate >> the use of Hadoop/HBase in commercial systems) then the companies >> businesses or >> products might be threatened. This means, unless something is done, >> companies >> cannot select Hadoop/HBase anymore for implementations because this is too >> much >> of a risk... which is in fact the biggest threat to hadoop... it was >> becoming >> popular and companies started to consider supporting it (providing >> development >> resources...) because this was a possible platform for their businesses... >> >> In my opinion a healthy attitude to this would be to analyse what was >> actually >> *really* protected and be creative on how hadoop could/should >> differentiate. >> Because if hadoop is just a 1:1 replica of the Google system, then there >> is no >> chance that it will attract more than the academic community and its nice >> success story is going to end here... I see enough intellectual power in >> the >> team to be able to take the Google patent and produce the next >> generation... >> >> My two cents, >> Laurence Hubert >> >> >> >> ----- Original Message ----- From: "Kay Kay" >> To: >> Sent: Thursday, January 21, 2010 3:28 AM >> Subject: Re: Google patent over Map Reduce - Hbase reflections >> >> >> > On 1/20/10 3:44 PM, stack wrote: >> >> I've been following the thread. I would tend to side with the general >> >> >> tenor >> >> that has it that its likely a just-in-case move by Google and that the >> >> likelihood of a Google suing Apache is not likely to happen in this >> >> dimension. >> >> >> >> >> > That was my general idea as well. >> > >> >> Are you (or your employer) spooked Kay Kay? >> >> >> >> >> > Not at all - but just started this to see what the opinions of the >> > community might be w.r.t. this. >> > >> >> St.Ack >> >> >> >> >> >> On Wed, Jan 20, 2010 at 3:34 PM, Kay Kay wrote: >> >> >> >> >> >>> A big thread currently going on at the hadoop common user mailing >>> >> >>> list - >> >>> >> >>> >> >>> http://mail-archives.apache.org/mod_mbox/hadoop-common-user/201001.mbox/< >> >>> 2c36b701001200817g77f245b1x6ba9d7d2cfd9e...@mail.gmail.com> . >> >>> >> >>> A good number of you might have already seen that thread, but just >>> >> >>> opening >> >>> up a thread for discussion to see what the thoughts of the community >> >>> >>> are , >> >>> w.r.t. patent and how much (if at all) of the application would be >>> >> >>> related >> >>> to that / any refactorings as necessary as seen by the team or >>> >> >>> thoughts in >> >>> general to the same. >> >>> >> >>> >> >>> >> >>> >> >> >> > >> > >> >> >> >> -------------------------------------------------------------------------------- >> >> >> >> Ce message entrant est certifié sans virus connu. >> Analyse effectuée par AVG - www.avg.fr >> Version: 9.0.730 / Base de données virale: 271.1.1/2635 - Date: 01/20/10 >> 20:18:00 > > > > > > > > > -------------------------------------------------------------------------------- > > > > Ce message entrant est certifié sans virus connu. > Analyse effectuée par AVG - www.avg.fr > Version: 9.0.730 / Base de données virale: 271.1.1/2636 - Date: 01/21/10 > 08:34:00 > >
