directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Emmanuel L├ęcharny <>
Subject Status...
Date Tue, 02 Apr 2013 17:53:52 GMT
Hi guys,

it was two tough weeks, with merely no work on the server... But still,
we made some progress on some side projects critical to ApacheDS. Let me
provide some headsup.

1) We are making progress on Mavibot. Mavibot is the backend replacement
for JDBM. Currently, we are able to manage the additions, deletions and
modifications in it. The performances are good, assuming the right page
size/nb of elements per page are correctly set (up to 13 000 updates per
second). The searches are blasting fast : 1 600 000 lookup per second.

Kiran has added support for multiValues in it, and is currently testing
a partition for mavibot. I let him giving some feedback on this part,
but I'm quite exited !

We have a lot to do, still. We currently never remove any revision, so
the file keeps growing after each modification. Kiran is working on a
bulk load mechanism. I'm working on a mechanism that reclaim the pages
that are not anymore in use.

We will have to add a way to access the old revisons too and I'm
currently working on that. This will allow us to get rid of the
ChangeLog system, as it will now be just a matter to switch to an old
revision to get back to the previous data set. That will make the test
way faster !

I expect to have Mavibot completed by the end of april.

2) MINA 3 is also on its way, and sounds promizing. It's already 50%
faster than MINA2, which is a huge gap. I hope to be able to build a
working prototype based on MINA 3 by the end of april too.

3) In the mean time, I think we can get a 2.0-RC1 released. We have
fixed some serious bugs lately, and I don't think we will add new
features in this version.

All in all, I think that once 2.0 will be out, we will be able to switch
to Mavibot backend and MINA 3, to get way better performances.

4) The documentation

It *absolutely* sucks :/ I have tried to improve the Kerberos
documentation last month, it was a lot of work, as the kerberos server
was also quite buggy. There is a lot to do, since we switched to the new
system last year, we haven't made a lot of progress in this area. This
is true for the server but also for the API. I can't seriously see how
we can get a 2.0-GA if the documentation is not improved in one way or
another... It's useless to have a fast and working server, if no one
know how to use it :/

That's pretty much it, Kiran, feel free to complete this mail.
Pierre-Arnaud, sorry if I haven't mentionned what you have done on
Studio, so please, feel free to add your progress !

Many thanks guys !

Emmanuel L├ęcharny 

View raw message