[Beowulf] Poll - Directory implementation

[Tom Harvill]

On 10/24/2018 12:44 PM, Michael Di Domenico wrote:
> On Wed, Oct 24, 2018 at 1:35 PM Ryan Novosielski <novosirj at rutgers.edu> wrote:
>> On 10/24/2018 01:30 PM, Ryan Novosielski wrote:
>>> On 10/24/2018 01:13 PM, Michael Di Domenico wrote:
>>>> On Wed, Oct 24, 2018 at 1:04 PM Ryan Novosielski
>>>> <novosirj at rutgers.edu> wrote:
>>>>> Funny, we are considering the exact opposite, and this is our
>>>>> motivation:
>>>>>
>>>>> https://access.redhat.com/solutions/2440481
>>>> we're contemplating the same, but we're okay with switching back
>>>> to the openldap source.  in my opinion redhat deprecating
>>>> openldap is just a money grab to push people towards rhds
>>> I don't have an alternate theory, but 389-ds is free, and I guess
>>> I imagined "comparable." I suppose if it's not as capable as
>>> OpenLDAP and RHDS is, that would make sense. But maybe they just
>>> want to push people toward something where there's an easy path?
>>>
>>> People saying that 389-ds is slow is not encouraging, given that
>>> we're currently attempting to tune OpenLDAP to be less slow.
> i don't want to diverge this thread from the OP, but how fast does
> ldap really need to be?  i have ~700 machines talking to two openldap
> servers w/ ssl enabled.  we have to run nslcd on the clients, but all
> is well
OP here - it's fine.

We run three 389-ds servers (multi-master) with clunky load-balancing 
across about the same number of machines using sssd. The LDAP servers 
are VMs on old hardware with 1GB RAM and don't break a sweat.