trafodion-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stack <st...@duboce.net>
Subject Re: [DISCUSSION] Work towards graduation
Date Tue, 21 Feb 2017 23:32:50 GMT
On Tue, Feb 21, 2017 at 12:03 PM, Rohit Jain <rohit.jain@esgyn.com> wrote:

> James,
>
> You bring up a very pertinent point that we brought up to Stack the last
> time.  There are projects that one could argue may have the same risk as
> Trafodion of not prevailing if a company does not survive.  But I guess
> that risk is mitigated somewhat to where those companies, such as Cloudera,
> are in their market presence versus Esgyn perhaps.  But the other reason
> provided was the involvement of many folks associated with Kudu, as an
> example, with other open source projects.  Our committers have not had as
> much involvement with other projects, perhaps because the complexity and
> the backlog of what we need to accomplish for our customers is large enough
> that it has not afforded us time to contribute towards other open source
> projects, even though we have always had the intent with HBase, ORC, etc.
>
> We have a good number of customers in China and a modest presence in the
> US.  But our customers so far themselves may not have the open source
> culture, or the resources to contribute to the project itself.  Plus, most
> of our code is in C++, although we have provided guidance to new developers
> on how to contribute towards the fair amount of code that we do have in
> Java.  So, certainly these have hindered the growth of the community.
>
> There is increasing frustration within Esgyn about open source and open
> sourcing anything into Apache since there is a huge cost to the company of
> maintaining an extra set of threads and versions, with no obvious path to
> TLP because of the reasons mentioned.  It seems that satisfying the
> decision makers, despite personal declarations of developers that they
> would be involved with the project beyond Esgyn, and that it would be crazy
> to think that no one would be interested in picking up such an incredible
> IP if anything were to happen to Esgyn -- this is decades of hundreds of
> million dollars of investment, into an incredible database technology,
> capable of running TPC-C and TPC-DS at very impressive numbers compared to
> the competition, with all queries executing while fully complying with the
> specs on the syntax (that no other vendor has been able to achieve in the
> Big Data world).  Full Hybrid Transactional/Analytical Processing support
> on Hadoop with unmatched performance on both ends of the spectrum.  Maybe
> we are just horrible at Marketing that a jewel of an engine like Trafodion
> must fight to get to TLP after all that we have done to try to make it
> ready for it.
>
> So, this is an ongoing struggle.  From what I understand all projects are
> supposed to have 2-3 mentors.  We have had Stack who has done a great job.
> But we need other strong mentors who can actively back the project and
> present its value to the Apache Foundation and what we have accomplished to
> qualify for TLP.  We have requested more mentors, but the same decision
> makers on TLP, seem to have ignored those requests.  So, go figure how the
> Apache foundation and its community works.
>
> I probably have stepped beyond the line in what I have said out of my own
> frustrations.  These in no way reflect the views of Esgyn but as an
> individual associated with Apache Trafodion, as they should.
>
>
Thanks Rohit for the straight perspective.

 * Yeah, the project needs more (active) mentors if we are going to be in
incubator for some time to come. We have close to zero though I'd consider
what Pierre does, 1st quality mentor-work (Interesting is that incubator is
doing a bit of house cleaning of deadbeat mentors like myself these times).
We can ask again for mentors upstream? We might do better this time around?
(James, you interested?)
 * Mentors will not bring users and contributors though; they just provide
guidance.
 * If there are two repositories, an internal and an external one (Apache),
then, yeah, mirroring and upkeep is a tax on a small companies' resources.
If no contrib coming in from outside, I can see that folks would ask why
bother.
 * All the issues apparent here while in incubator will be present when you
TLP; TLPing is not a project entering project paradise.
 * Apache, and incubator particularly, is about successful communities
first. Many successful Apache projects are grounded upon tech that is dodgy
even, nowhere close to the jewel that Trafodion is.

If folks here think we are ready, we can go for a vote to graduate but as
is Pierre's experience, we'll be quizzed on "...whether the PPMC is diverse
enough to be regarded as independent of any outside affiliation." While we
have companies other than Esgyn listed as affiliations of PPMC members, how
many are active? One or two would be enough I'd think. Do we have any users
who'd be happy to talk up their use of Apache Trafodion publicly.

Thanks,
S





> Rohit
>
> -----Original Message-----
> From: James Taylor [mailto:jamestaylor@apache.org]
> Sent: Tuesday, February 21, 2017 12:28 PM
> To: dev@trafodion.incubator.apache.org
> Subject: Re: [DISCUSSION] Work towards graduation
>
> On Fri, Feb 17, 2017 at 4:55 PM, Stack <stack@duboce.net> wrote:
>
> >
> > Is there general agreement with Pierre's belief? If there were no Esgyn,
> > would Trafodion prevail? Asking here so we are prepared when question
> > comes up on the general incubator list.
> >
> >
> Did that question come up wrt Cloudera and Kudu for graduation?
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message