[ENet-discuss] Reliable packets and data sending approaches

Chris Jurney jurney at gmail.com
Thu Oct 29 14:20:03 PDT 2009 

50 is really high, but if your packets are really small, it's possibly
maintainable.  I'd say setup a game with your maximum player count/bandwidth
usage, and measure your bandwidth with any network monitoring tool (task
manager will even give you a rough idea).  Keep all the users up channels
below 128k**bit** with a little margin, and you should be fine.

Chris

On Wed, Oct 28, 2009 at 1:25 PM, Philip Bennefall
<philip at pbsoundscape.net>wrote:

>  Thank you Chris, for your very helpful information. based on this, do you
> think that perhaps 50 would be a reasonable maximum?
>
> I will do tests on a whireless connection that I've got here to get a rough
> idea of packet loss and round trip times and so on, but of course that's
> only one configuration and is not going to represent an average by any
> stretch of the imagination, but I'll fiddle around with rates and see where
> it gets me.
>
> Again, thank you!
>
> Kind regards,
>
> Philip Bennefall
>
> ----- Original Message -----
> *From:* Chris Jurney <jurney at gmail.com>
> *To:* Discussion of the ENet library <enet-discuss at cubik.org>
> *Sent:* Wednesday, October 28, 2009 8:10 PM
> *Subject:* Re: [ENet-discuss] Reliable packets and data sending approaches
>
> When you're picking your update rate, keep in mind users' up channel
> limitations.  128kbit is a very common cap in Internetland.  I think the
> size of an unreliable eNet header (~32 bytes) + UDP (8 bytes) + IP (20
> bytes) gives you a minimum packet size of roughly 60 bytes.
>
> Upstream header overhead = 60 byte header * rate * 8 bits/packet
>
> If you send at 60/s, you'll have at least 29kbit of packet overhead before
> you send your first byte of payload.  If you're on a console, that overhead
> potentially goes up with their wrapper as well.
>
> (I'm not 100% sure of my size number for eNet because we have fiddled with
> headers a bit)
>
> Chris
>
> ----- Original Message ----- From: "Lee Salzman" <lsalzman1 at cox.net>
>
>> To: "Discussion of the ENet library" <enet-discuss at cubik.org>
>> Sent: Tuesday, October 27, 2009 2:27 PM
>>
>> Subject: Re: [ENet-discuss] Reliable packets and data sending approaches
>>
>>
>> Don't rely on the throttle. Choose a reasonable rate to begin with.
>>> 20-30 times a second is probably fair. Keep in mind that on average an
>>> event will occur half-way between an interval, so 20 Hz does not
>>> correspond to 50 ms latency, but rather on average more like 25 ms, and
>>> by the time you get to 30 Hz your average latency is like 16 ms. Taking
>>> that up to 50 Hz, and your average latency is only about 10 ms, so
>>> you're making huge jumps in bandwidth usage for very marginal benefits.
>>>
>>> Lee
>>>
>>> Philip Bennefall wrote:
>>>
>>>> I understand what you're saying there. But say then that I start at a
>>>> rate of 50 per second, and then let ENet's dynamic throttle take it
>>>> down if necessary? Would that be a safe approach? It would allow for
>>>> 50 packets a second in ideal network conditions such as a lan or two
>>>> super connections, and automatically adapt itself to other
>>>> circumstances. What do you think?
>>>>
>>>> Kind regards,
>>>>
>>>> Philip Bennefall
>>>>
>>>>    ----- Original Message -----
>>>>    *From:* Nuno Silva <mailto:little.coding.fox at gmail.com>
>>>>    *To:* Discussion of the ENet library <mailto:enet-discuss at cubik.org>
>>>>    *Sent:* Tuesday, October 27, 2009 8:04 AM
>>>>    *Subject:* Re: [ENet-discuss] Reliable packets and data sending
>>>>    approaches
>>>>
>>>>    60 times per second would probably be overkill on most
>>>>    connections, considering you send packets every 16ms, which IMHO
>>>>    may be a bit too fast even for TCP. Do notice that i'm no
>>>>    networking expert, but having a guy from the other side of the
>>>>    world send/receive packets every 16ms instead of the usual 50ms
>>>>    will need a pretty darn good connection.
>>>>
>>>>    On Tue, Oct 27, 2009 at 3:47 AM, Philip Bennefall
>>>>    <philip at pbsoundscape.net <mailto:philip at pbsoundscape.net>> wrote:
>>>>
>>>>        Lee,
>>>>
>>>>        Would it be acceptable to send small packets out, say 60 times
>>>>        a second or so? Will ENet handle it if it getst oo much?
>>>>
>>>>        Kind regards,
>>>>
>>>>        Philip Bennefall
>>>>        ----- Original Message ----- From: "Lee Salzman"
>>>>        <lsalzman1 at cox.net <mailto:lsalzman1 at cox.net>>
>>>>        To: "Discussion of the ENet library" <enet-discuss at cubik.org
>>>>        <mailto:enet-discuss at cubik.org>>
>>>>        Sent: Tuesday, October 27, 2009 4:00 AM
>>>>
>>>>        Subject: Re: [ENet-discuss] Reliable packets and data sending
>>>>        approaches
>>>>
>>>>
>>>>            Mihai is mistaken. Sauerbraten only sends 30 times a
>>>>            second. Events like
>>>>            gun shots are sent reliably. Only position data for
>>>>            players is sent
>>>>            unreliably.
>>>>
>>>>            Lee
>>>>
>>>>            Philip Bennefall wrote:
>>>>
>>>>                So what is the game frame rate in sauerbraten? How
>>>>                often does it end
>>>>                up sending updates, how many times a second?
>>>>
>>>>                Kind regards,
>>>>
>>>>                Philip Bennefall
>>>>
>>>>
>>>>            _______________________________________________
>>>>            ENet-discuss mailing list
>>>>            ENet-discuss at cubik.org <mailto:ENet-discuss at cubik.org>
>>>>            http://lists.cubik.org/mailman/listinfo/enet-discuss
>>>>
>>>>
>>>>
>>>>
>>>>  --------------------------------------------------------------------------------
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>        No virus found in this incoming message.
>>>>        Checked by AVG - www.avg.com <http://www.avg.com>
>>>>        Version: 8.5.423 / Virus Database: 270.14.32/2459 - Release
>>>>        Date: 10/25/09 19:57:00
>>>>
>>>>        _______________________________________________
>>>>        ENet-discuss mailing list
>>>>        ENet-discuss at cubik.org <mailto:ENet-discuss at cubik.org>
>>>>        http://lists.cubik.org/mailman/listinfo/enet-discuss
>>>>
>>>>
>>>>
>>>>  ------------------------------------------------------------------------
>>>>    _______________________________________________
>>>>    ENet-discuss mailing list
>>>>    ENet-discuss at cubik.org
>>>>    http://lists.cubik.org/mailman/listinfo/enet-discuss
>>>>
>>>>
>>>>  ------------------------------------------------------------------------
>>>>
>>>>    No virus found in this incoming message.
>>>>    Checked by AVG - www.avg.com
>>>>    Version: 8.5.423 / Virus Database: 270.14.33/2461 - Release Date:
>>>>    10/26/09 20:22:00
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> _______________________________________________
>>>> ENet-discuss mailing list
>>>> ENet-discuss at cubik.org
>>>> http://lists.cubik.org/mailman/listinfo/enet-discuss
>>>>
>>>>
>>> _______________________________________________
>>> ENet-discuss mailing list
>>> ENet-discuss at cubik.org
>>> http://lists.cubik.org/mailman/listinfo/enet-discuss
>>>
>>
>>
>> --------------------------------------------------------------------------------
>>
>>
>>
>>
>>
>> No virus found in this incoming message.
>> Checked by AVG - www.avg.com
>> Version: 8.5.423 / Virus Database: 270.14.33/2461 - Release Date: 10/26/09
>> 20:22:00
>>
>> _______________________________________________
>> ENet-discuss mailing list
>> ENet-discuss at cubik.org
>> http://lists.cubik.org/mailman/listinfo/enet-discuss
>>
>
>  ------------------------------
>
> _______________________________________________
> ENet-discuss mailing list
> ENet-discuss at cubik.org
> http://lists.cubik.org/mailman/listinfo/enet-discuss
>
> ------------------------------
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 8.5.423 / Virus Database: 270.14.34/2463 - Release Date: 10/27/09
> 15:50:00
>
>
> _______________________________________________
> ENet-discuss mailing list
> ENet-discuss at cubik.org
> http://lists.cubik.org/mailman/listinfo/enet-discuss
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.cubik.org/pipermail/enet-discuss/attachments/20091029/dcc7d551/attachment.htm>

More information about the ENet-discuss mailing list