(Apologies; I'm on the way to a family memorial service, so I don't have
time for a complete answer here, although this is an important issue for
trademarks@)
I certainly appreciate the work that the Hadoop PMC has been doing in
terms of better defining the kinds of use cases for Hadoop derivatives
or related kinds of software! The pre-existing "Powered By" metaphor
from httpd is a great model, and one that allows for strong independent
branding by third parties while still allowing for use of Apache marks
within third party names, with a clear separation.
However the phrase "Powered by" does not fit many uses of software like
Hadoop, so we need both some alternate naming styles, as well as some
better definitions of models of third party uses; derivatives with
additional code, integrations with other products, uses of Apache
software as services, etc. I've always wanted to create a generic list
of other phrases to use instead of "Powered By" that we would list as a
policy of pre-approved uses for these kind of cases.
I'd like to see more discussion from the PMC, because Hadoop in
particular is a key technology that's being used in a wide variety of
ways - really defining a new computing model. However I think we really
need to take this input and then have trademarks@ work on an Apache-wide
policy for approved third party product and services naming styles for
all Apache products. Obviously, some of the specific types of uses of
Hadoop may not be applicable to other Apache products, but in general I
think working on a larger set of guidelines that apply would be the best
way to approach this situation for all Apache projects.
Overall, trademarks@ and VP, Brand Management set branding policy for
all Apache projects. But we definitely need the combined input of both
folks on Hadoop - with so many new and innovative uses of our software -
plus the input of some of our longer-running projects like httpd and
Tomcat, with experience in *nix-style packaging and the like, to come up
with the best policy for these kind of third party uses of our product
marks.
- Shane Curcuru
VP, Brand Management, The Apache Software Foundation
Segel, Mike wrote:
Owen, From your response below you say the following: "We are trying
to assert that only the Apache releases can be called Hadoop. That
seems to be the best way to help the project ensure compatibility and
prevent user confusion." and " I want Hadoop to be used in as many
products as possible. Having a FooCo product that is called "FooCo
HugeInsights powered by Hadoop" is absolutely great. The question is
just whether they can call something Hadoop if it isn't an Apache
release.
-- Owen"
Owen, Unfortunately what you're saying is that you would only approve
of companies that build their products on top of Apache's release and
doesn't modify the Apache release. To give you an example... if Acme
Risk Management Company sold a product using Hadoop to do risk
analysis on a bank's portfolio, they can only say "powered by Hadoop"
if they build their application on top of Apache's release. But the
minute they build their solution on top of anyone else, they would
lose that right? So using Cloudera's release, which contains things
outside of the official Apache release would disallow them? Or if
they make their own modifications to the underlying release which
isn't part of the official release, they could no longer make that
claim?
This interpretation of "powered by Hadoop" would unfortunately lead
to as many problems as it attempts to solve. First, many choose
Cloudera's release because they sell commercial support. So in
choosing Cloudera's release, they would lose the ability to say
"powered by Hadoop". This diminishes the branding message.
The Apache License allows for broad reuse and relicensing as long as
the company complies with Apache's T's & C's. Limiting the ability
to say "powered by Hadoop" means that they will say that their
solution uses a commercially supported 'derivative of Hadoop'. In
terms of legalese, good luck in trying to get them on a misuse of
your trademark. Cloudera, EMC, MapRTech, Datastax all offer
derivatives of Hadoop. (I'm not forgetting about Yahoo!, but are they
releasing their own version as well?) The term Hadoop is used as a
reference to Apache's Hadoop release.
I hope that you start to see the dangers on taking a narrow approach
in how you define Hadoop.
Just IMHO.
-Mike
-----Original Message----- From: Owen O'Malley
[mailto:[email protected]] Sent: Thursday, June 16, 2011 1:32 PM To:
[email protected] Cc: [email protected] Subject: Re:
Trademarks and Derivative Works
On Jun 16, 2011, at 10:59 AM, Lawrence Rosen wrote:
Under default trademark law, those others can distribute HADOOP and
APACHE HADOOP only if it is a redistribution of *our* HADOOP or
APACHE HADOOP software. (That's why you can buy Jello Brand gelatin
at Safeway.) Those trademarks are our names for our software. ASF
is the source and origin of those software goods. Nobody else can
apply those trademarks to their own software.
The problem is that a rapidly growing set of companies are
distributing products that have never been released by Apache and
calling them Hadoop. The rules from HTTPD, as I understand them, are
that they allow artifacts to be called HTTPD that are releases plus
patches that have been committed. With HTTPD that has a formal
specification and a very large compatibility test suite, that works.
For Hadoop without a formal specification or test suite, we simply
can't handle companies calling things Hadoop that are thousands of
patches away from our releases. We are trying to assert that only the
Apache releases can be called Hadoop. That seems to be the best way
to help the project ensure compatibility and prevent user confusion.
It will be to our advantage to have HADOOP and APACHE HADOOP
software better known and widely used throughout the world. For
that purpose, we should be defining the rules we want to
*encourage* third parties to follow, not arguing about derivative
work analysis or voting on whether or not something is a trademark.
I want Hadoop to be used in as many products as possible. Having a
FooCo product that is called "FooCo HugeInsights powered by Hadoop"
is absolutely great. The question is just whether they can call
something Hadoop if it isn't an Apache release.
-- Owen
The information contained in this communication may be CONFIDENTIAL
and is intended only for the use of the recipient(s) named above. If
you are not the intended recipient, you are hereby notified that any
dissemination, distribution, or copying of this communication, or any
of its contents, is strictly prohibited. If you have received this
communication in error, please notify the sender and delete/destroy
the original message and any copy of it from your computer or paper
files.
