directory-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kiran Ayyagari <kayyag...@apache.org>
Subject Re: Evaluating ApacheDS as part of a preemptive cache
Date Tue, 26 Feb 2013 22:21:30 GMT
On Wed, Feb 27, 2013 at 3:42 AM, Emmanuel Lécharny <elecharny@gmail.com>wrote:

> Le 2/26/13 8:39 PM, Tom Bruser a écrit :
> > -----Original Message-----
> > From: ayyagarikiran@gmail.com [mailto:ayyagarikiran@gmail.com] On
> Behalf Of
> > Kiran Ayyagari
> > Sent: Tuesday, February 26, 2013 1:36 PM
> > To: users@directory.apache.org
> > Subject: Re: Evaluating ApacheDS as part of a preemptive cache
> >
> >>> On Tue, Feb 26, 2013 at 11:54 PM, Tom <ml@tombruser.com> wrote:
> >>> -----Original Message-----
> >>> From: ayyagarikiran@gmail.com [mailto:ayyagarikiran@gmail.com] On
> >>> Behalf Of Kiran Ayyagari
> >>> Sent: Tuesday, February 26, 2013 2:50 AM
> >>> To: users@directory.apache.org
> >>> Subject: Re: Evaluating ApacheDS as part of a preemptive cache
> >>>
> >>> On Tue, Feb 26, 2013 at 12:58 PM, Tom <ml@tombruser.com> wrote:
> >>>
> >> I would suggest you to go with the 2.0.0-M10(or M11 which will be
> >> released
> >> next week)
> >> The current 2.0 (milestone) version is much more advanced and feature
> rich
> > than the 1.5.x series
> >
> > I'm surprised to see this answer since from what I understand, GA hasn't
> > been scheduled for 2.0 yet.
>
> GA is just a convention. The Apache Software Foundation moto is 'release
> fast, release often', so even if we go for a GA, we will most certainly
> have another release a few weeks/months after.
>
> FTR, we have produced 11 releases for 2.0.0 in less than 2 years, that's
> around a release every two months. Regarding the 1.5.7 version, it's
> nearly 3 years old. 7940 commits have been done (out of 25783 since the
> beginning of the project 8 years ago), so you can consider that 30% of
> the code has ben changed : this is far from being minor...
>
> But enough with the numbers : 1.5.7 was never considered as a "stable"
> version. Here, "Stable" means the API is stable. In any case, every
> version is supposed to be kind of 'good enough' to be used, except that
> we always consider that the latest is the best one - until the next one.
>
> WRT 2.0 version, we consider that the current milestones are probably
> the last ones before we move to a RC1 (we may have one or two more
> milestones before a RC1). We expect those milestones to be released in
> the next 2 months.
>
>
> > I apologize as I should have mentioned that
> > this is for a mission critical system with up to a 250K user base if it
> > makes a difference.  Is the current milestone release still the
> > recommendation based on that?
> Yes. Even if the current version will require a lot of time to load
> those 250K entries (at 200 entries per second, count 1h more or less,
>
just to clarify a bit, this is the time to *add* new entries.
Search operation in embedded mode returns 79000+ results per second

> depending on the number of indexes you will have), it will deliver way
> better performances than 1.5.7, but more important, it will be more
> stable, and has less bugs.
>
> One example : beside the replication that has been added 6 months ago,
> we have now a mechanism that guarantee that the server can be restarted
> even in case of a brutal crash. Concurrent modifications can be done -
> something that failed in 1.5.7. And many other improvements...
>
> Let face it : it's not perfect. It's not finished. But no piece of
> software like that is perfect nor finished. We are doing our best to get
> it better, that your guarantee :-)
>
> Hoep it helps...
>
> --
> Regards,
> Cordialement,
> Emmanuel Lécharny
> www.iktek.com
>
>


-- 
Kiran Ayyagari
http://keydap.com

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