Send kea-dev mailing list submissions to
[email protected]
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.isc.org/mailman/listinfo/kea-dev
or, via email, send a message with subject or body 'help' to
[email protected]
You can reach the person managing the list at
[email protected]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of kea-dev digest..."
Today's Topics:
1. Re: kea apache cassandra/datastax cassandra (Razvan Becheriu)
----------------------------------------------------------------------
Message: 1
Date: Thu, 19 May 2016 17:44:40 +0300
From: Razvan Becheriu <[email protected]>
To: Angelo Failla <[email protected]>
Cc: "[email protected]" <[email protected]>
Subject: Re: [kea-dev] kea apache cassandra/datastax cassandra
Message-ID:
<CAPj=nfus-dqfcsx6vrvfehfkcda36xccvuzfusywtulnmhu...@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Hi,
I did the implementation directly in the KEA server. I did not use any
hooks for the cassandra backend support.
I did not perform stress tests on the cassandra implementation.
Regards,
Razvan
On Thu, May 19, 2016 at 5:21 PM, Angelo Failla <[email protected]> wrote:
> Nice!
>
>
>
> Have you stress tested your patch with perfdhcp with, say, 1500 RPS?
>
>
>
> As I wrote in a previous thread [1] KEA is a mono threaded application and
> your hook might not perform as expected if your backend is slow (e.g.
> between 20ms and 50ms).
>
> Even if your backend takes 8ms to respond you will probably still max out
> at 125 RPS (best conditions).
>
> This is because when a request is being served all the incoming udp
> packets will be buffered in kernel space, and processed sequentially? and
> if your hook is not fast enough your network recvq will max out until it
> hits the kernel buffer limits, after which you will start dropping packets.
>
>
>
> You might need to alternatives solutions to this. For example the solution
> we use in Facebook (in our hook) is to drop the first DISCOVERY/SOLICIT
> request, spawn a background fetch to our backend, which populate internal
> data structure in memory, when the client retries a second or so later the
> info will be in memory and therefore it can be served fast. We basically
> take advantage of the fact that clients usually have retry/exponential
> backoff logic.
>
>
>
> If you run into performance issues (like we did) you might find my
> proposal viable (until and if KEA people will consider redesign the core to
> be async or multithread).
>
>
>
> [1] https://lists.isc.org/pipermail/kea-dev/2016-March/000628.html
>
> --
>
> Angelo Failla
>
> Cluster Infrastructure - Dublin
>
> [email protected]
>
>
>
> *From: *<[email protected]> on behalf of Razvan Becheriu <
> [email protected]>
> *Date: *Thursday, May 19, 2016 at 9:58 AM
> *To: *"[email protected]" <[email protected]>
> *Subject: *[kea-dev] kea apache cassandra/datastax cassandra
>
>
>
> Hi,
>
>
>
> My name is Razvan Becheriu.
>
>
>
> I work for Qualitance in collaboration with Deutsche Telekom.
>
> I have been working on improving the KEA server by adding support for
> apache cassandra/datastax cassandra.
>
> From my understanding, the new source files will be commited with "Apache
> License".
>
>
>
> I am ready to commit the changes to a separate branch (for code review).
> The source files have been updated to latest (main branch) version.
>
>
>
> One thing to mention is that not all unit tests have been written for the
> cassandra backend.
>
>
>
> Please let me know how to proceed.
>
>
>
> Regards,
>
> Razvan
>
>
>
>
>
>
> --
>
> *Razvan Becheriu* | senior software engineer | QUALITANCE?
> *RO:* +00 (000) 000 000 | *office:* +40 (372) 944 741
> Opera Center bldg. 2, fl. 7-8, 2 Dr. Nicolae D. Staicovici, Bucharest
> 050556
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__www.google.ro_maps_place_QUALITANCE_-4044.433653-2C26.079464-2C17z_data-3D-214m6-211m3-213m2-211s0x40b1ff173c5bc0b9-3A0x7226343f810c7ede-212sQUALITANCE-213m1-211s0x40b1ff173c5bc0b9-3A0x7226343f810c7ede-3Fhl-3Dro&d=CwMFaQ&c=5VD0RTtNlTh3ycd41b3MUw&r=XQHjLQh2eTTW8a0c8kaZ9Q&m=4FJSvXdaJAhK6dHo-ycbobR-0O43b8wsYhIS_w3ZsvA&s=Gg2OEEGB_Larq_eTb8TzCz6ZseeMou69mVoauRLWN5M&e=>
> [email protected] | www.qualitance.com
> <https://urldefense.proofpoint.com/v2/url?u=https-3A__www.qualitance.com_&d=CwMFaQ&c=5VD0RTtNlTh3ycd41b3MUw&r=XQHjLQh2eTTW8a0c8kaZ9Q&m=4FJSvXdaJAhK6dHo-ycbobR-0O43b8wsYhIS_w3ZsvA&s=NgDyDRXU1fCXOCOXNGGfK4tT9pJHog4M4lWDjpw_GPk&e=>
>
>
>
> --
> The information contained in this email is confidential. It is intended
> only for the stated addressee and access to it by any other person is
> unauthorised. If you are not an addressee, you must not disclose, copy,
> circulate or in another way use or reply to the information contained in
> this email. Such unauthorised use may be unlawful. If you have received
> this email in error, please notify the sender immediately and delete all
> copies from your computer and network.
>
>
>
--
*Razvan Becheriu* | senior software engineer | QUALITANCE?
*RO:* +00 (000) 000 000 | *office:* +40 (372) 944 741
Opera Center bldg. 2, fl. 7-8, 2 Dr. Nicolae D. Staicovici, Bucharest 050556
<https://www.google.ro/maps/place/QUALITANCE/@44.433653,26.079464,17z/data=!4m6!1m3!3m2!1s0x40b1ff173c5bc0b9:0x7226343f810c7ede!2sQUALITANCE!3m1!1s0x40b1ff173c5bc0b9:0x7226343f810c7ede?hl=ro>
[email protected] | www.qualitance.com
--
The information contained in this email is confidential. It is intended
only for the stated addressee and access to it by any other person is
unauthorised. If you are not an addressee, you must not disclose, copy,
circulate or in another way use or reply to the information contained in
this email. Such unauthorised use may be unlawful. If you have received
this email in error, please notify the sender immediately and delete all
copies from your computer and network.
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<https://lists.isc.org/pipermail/kea-dev/attachments/20160519/d329a894/attachment-0001.html>
------------------------------
_______________________________________________
kea-dev mailing list
[email protected]
https://lists.isc.org/mailman/listinfo/kea-dev
End of kea-dev Digest, Vol 26, Issue 3
**************************************
