<div dir="auto">We don't use Stxxl during the query phase, but an SSD is recommended. We use an external memory storage for the R-Tree that we use for getting the closed street segment. <div dir="auto"><br></div><div dir="auto">For matrix queries on MLD are about 2-3 times slower than on CH.</div><div dir="auto"><br></div><div dir="auto">Best, </div><div dir="auto">Patrick </div></div><br><div class="gmail_quote"><div dir="ltr">Baig, Tariq <<a href="mailto:tariq.baig@ds.mpg.de">tariq.baig@ds.mpg.de</a>> schrieb am Do., 29. März 2018, 11:27:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Thanks for the suggestions Patrick,<br>
<br>
how do you think things would change if we needed trafic support? I guess MLD is slower than CH,<br>
could we compensate for that by putting the whole system on an SSD to make the stxxl perform better?<br>
<br>
Thanks again,<br>
tariq<br>
----------------------------------------------------------------------------------<br>
<br>
Tariq Baig-Meininghaus<br>
<br>
Researcher<br>
<br>
Max Planck Institute for Dynamics and Self-Organization<br>
Department DCF/EcoBus<br>
<br>
Am Faßberg 17<br>
<br>
37077 GÖTTINGEN<br>
GERMANY<br>
<br>
<br>
Mail: <a href="mailto:tariq.baig@ds.mpg.de" target="_blank" rel="noreferrer">tariq.baig@ds.mpg.de</a><br>
<br>
Web: <a href="http://www.ds.mpg.de" rel="noreferrer noreferrer" target="_blank">www.ds.mpg.de</a>, <a href="http://ecobus.fokos.info" rel="noreferrer noreferrer" target="_blank">ecobus.fokos.info</a><br>
<br>
________________________________________<br>
From: <a href="mailto:osrm-talk-request@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk-request@openstreetmap.org</a> [<a href="mailto:osrm-talk-request@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk-request@openstreetmap.org</a>]<br>
Sent: Wednesday, March 28, 2018 2:00 PM<br>
To: <a href="mailto:osrm-talk@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk@openstreetmap.org</a><br>
Subject: OSRM-talk Digest, Vol 63, Issue 11<br>
<br>
Send OSRM-talk mailing list submissions to<br>
<a href="mailto:osrm-talk@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk@openstreetmap.org</a><br>
<br>
To subscribe or unsubscribe via the World Wide Web, visit<br>
<a href="https://lists.openstreetmap.org/listinfo/osrm-talk" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/osrm-talk</a><br>
or, via email, send a message with subject or body 'help' to<br>
<a href="mailto:osrm-talk-request@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk-request@openstreetmap.org</a><br>
<br>
You can reach the person managing the list at<br>
<a href="mailto:osrm-talk-owner@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk-owner@openstreetmap.org</a><br>
<br>
When replying, please edit your Subject line so it is more specific<br>
than "Re: Contents of OSRM-talk digest..."<br>
<br>
<br>
Today's Topics:<br>
<br>
1. Suggested hardware specs for server under heavy use (Baig, Tariq)<br>
2. Re: Suggested hardware specs for server under heavy use<br>
(Patrick Niklaus)<br>
3. Re: OSRM 5.14.3 set kilometers long ways as small elements<br>
(Daniel Patterson)<br>
<br>
<br>
----------------------------------------------------------------------<br>
<br>
Message: 1<br>
Date: Tue, 27 Mar 2018 12:49:08 +0000<br>
From: "Baig, Tariq" <<a href="mailto:tariq.baig@ds.mpg.de" target="_blank" rel="noreferrer">tariq.baig@ds.mpg.de</a>><br>
To: "<a href="mailto:osrm-talk@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk@openstreetmap.org</a>" <<a href="mailto:osrm-talk@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk@openstreetmap.org</a>><br>
Subject: [OSRM-talk] Suggested hardware specs for server under heavy<br>
use<br>
Message-ID:<br>
<<a href="mailto:350C9BAAF4E5894AA7907627B5508828D532F4E8@UM-EXCDAG-A03.um.gwdg.de" target="_blank" rel="noreferrer">350C9BAAF4E5894AA7907627B5508828D532F4E8@UM-EXCDAG-A03.um.gwdg.de</a>><br>
Content-Type: text/plain; charset="iso-8859-1"<br>
<br>
Dear all,<br>
<br>
we are currently working on a research project that uses osrm as a routing backend.<br>
Since we heavily rely on OSRM we would like to migrate it to a more appropriate harware setup.<br>
<br>
What would be the recommended specs for a server that provides routing for the whole of germany,<br>
and can service order of 100 - 200 table requests per second?<br>
<br>
Greetings and thanks in advance,<br>
<br>
tariq<br>
<br>
----------------------------------------------------------------------------------<br>
<br>
Tariq Baig-Meininghaus<br>
<br>
Researcher<br>
<br>
Max Planck Institute for Dynamics and Self-Organization<br>
Department DCF/EcoBus<br>
<br>
Am Faßberg 17<br>
<br>
37077 GÖTTINGEN<br>
GERMANY<br>
<br>
<br>
Mail: <a href="mailto:tariq.baig@ds.mpg.de" target="_blank" rel="noreferrer">tariq.baig@ds.mpg.de</a><br>
<br>
Web: <a href="http://www.ds.mpg.de" rel="noreferrer noreferrer" target="_blank">www.ds.mpg.de</a>, <a href="http://ecobus.fokos.info" rel="noreferrer noreferrer" target="_blank">ecobus.fokos.info</a><br>
<br>
<br>
<br>
------------------------------<br>
<br>
Message: 2<br>
Date: Tue, 27 Mar 2018 14:41:18 +0000<br>
From: Patrick Niklaus <<a href="mailto:patrick.niklaus@student.kit.edu" target="_blank" rel="noreferrer">patrick.niklaus@student.kit.edu</a>><br>
To: Mailing list to discuss Project OSRM <<a href="mailto:osrm-talk@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk@openstreetmap.org</a>><br>
Subject: Re: [OSRM-talk] Suggested hardware specs for server under<br>
heavy use<br>
Message-ID:<br>
<<a href="mailto:CAFxP%2B1P1OFr6_ea6pcVifLau8pJqp5f4V4KWAdZ9gfno-gQ8Bw@mail.gmail.com" target="_blank" rel="noreferrer">CAFxP+1P1OFr6_ea6pcVifLau8pJqp5f4V4KWAdZ9gfno-gQ8Bw@mail.gmail.com</a>><br>
Content-Type: text/plain; charset="UTF-8"<br>
<br>
Hey Tariq,<br>
<br>
This heavily depends on the settings you use and the kind of requests.<br>
For the car profile Germany should fit into 16GB of RAM, but I would<br>
calculate some buffer depending on if you need the option to exclude<br>
motorways and ferries.<br>
I'll assume you don't need traffic support? Then using or Contraction<br>
Hierarchy toolchain might be your best option.<br>
<br>
For the CPU requirement to get close to 200 req/sec it entirely<br>
depends on the kind of request (how many input coordinates).<br>
For a 100x100 matrix you can expect something like ~100ms on Germany,<br>
that would put you at 10 req per core per second.<br>
A 16 core system should probably get you in the right order of magnitude.<br>
<br>
Best,<br>
Patrick<br>
<br>
<br>
On Tue, Mar 27, 2018 at 12:49 PM, Baig, Tariq <<a href="mailto:tariq.baig@ds.mpg.de" target="_blank" rel="noreferrer">tariq.baig@ds.mpg.de</a>> wrote:<br>
> Dear all,<br>
><br>
> we are currently working on a research project that uses osrm as a routing backend.<br>
> Since we heavily rely on OSRM we would like to migrate it to a more appropriate harware setup.<br>
><br>
> What would be the recommended specs for a server that provides routing for the whole of germany,<br>
> and can service order of 100 - 200 table requests per second?<br>
><br>
> Greetings and thanks in advance,<br>
><br>
> tariq<br>
><br>
> ----------------------------------------------------------------------------------<br>
><br>
> Tariq Baig-Meininghaus<br>
><br>
> Researcher<br>
><br>
> Max Planck Institute for Dynamics and Self-Organization<br>
> Department DCF/EcoBus<br>
><br>
> Am Faßberg 17<br>
><br>
> 37077 GÖTTINGEN<br>
> GERMANY<br>
><br>
><br>
> Mail: <a href="mailto:tariq.baig@ds.mpg.de" target="_blank" rel="noreferrer">tariq.baig@ds.mpg.de</a><br>
><br>
> Web: <a href="http://www.ds.mpg.de" rel="noreferrer noreferrer" target="_blank">www.ds.mpg.de</a>, <a href="http://ecobus.fokos.info" rel="noreferrer noreferrer" target="_blank">ecobus.fokos.info</a><br>
><br>
> _______________________________________________<br>
> OSRM-talk mailing list<br>
> <a href="mailto:OSRM-talk@openstreetmap.org" target="_blank" rel="noreferrer">OSRM-talk@openstreetmap.org</a><br>
> <a href="https://lists.openstreetmap.org/listinfo/osrm-talk" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/osrm-talk</a><br>
<br>
<br>
<br>
------------------------------<br>
<br>
Message: 3<br>
Date: Tue, 27 Mar 2018 08:10:20 -0700<br>
From: Daniel Patterson <<a href="mailto:daniel@mapbox.com" target="_blank" rel="noreferrer">daniel@mapbox.com</a>><br>
To: Mailing list to discuss Project OSRM <<a href="mailto:osrm-talk@openstreetmap.org" target="_blank" rel="noreferrer">osrm-talk@openstreetmap.org</a>><br>
Subject: Re: [OSRM-talk] OSRM 5.14.3 set kilometers long ways as small<br>
elements<br>
Message-ID:<br>
<<a href="mailto:CANvddSK075Y44OcGTMXu8bqdCW1aj62HPgjHULG3pZe%2BwRGM9Q@mail.gmail.com" target="_blank" rel="noreferrer">CANvddSK075Y44OcGTMXu8bqdCW1aj62HPgjHULG3pZe+wRGM9Q@mail.gmail.com</a>><br>
Content-Type: text/plain; charset="utf-8"<br>
<br>
Hi François,<br>
<br>
> I don't understand why flagging part of the graph as "small element"<br>
prevent a route to be found.<br>
<br>
I guess you could describe this as a bug. What OSRM is trying to do is<br>
always return a route. The current approach makes sure to snap the input<br>
coordinates to edges on the same "strongly connected component", which<br>
(usually) guarantees a route will be found.<br>
<br>
There's an older blog post by Denis here:<br>
<a href="https://blog.mapbox.com/robust-navigation-with-smart-nearest-neighbor-search-dbc1f6218be8" rel="noreferrer noreferrer" target="_blank">https://blog.mapbox.com/robust-navigation-with-smart-nearest-neighbor-search-dbc1f6218be8</a><br>
that explains the behaviour/problem that this snapping is intended to<br>
solve.<br>
<br>
This behaviour makes sense for fully disconnected components - islands,<br>
parks, etc.<br>
<br>
However, in your case, the components are weakly connected - there is a<br>
one-way path in, and an A->B route could be found, but a B->A route would<br>
not.<br>
<br>
This situation isn't really handled by the current approach. The Tarjan<br>
SCC algorithm flags *strongly connected* graph regions (full connectivity<br>
in both directions). We would need to do something quite different I think<br>
in order to do correct snapping for weakly connected components as you have<br>
in your graph.<br>
<br>
Generally in OSM, there aren't too many of these, and where they are,<br>
they're usually a mapping error (one-way roads with no exits, etc). Given<br>
that, I'm not sure it's worth spending time improving this behaviour for<br>
use with OSM data.<br>
<br>
daniel<br>
<br>
<br>
On Tue, Mar 27, 2018 at 1:58 AM, François Lacombe <<a href="mailto:fl.infosreseaux@gmail.com" target="_blank" rel="noreferrer">fl.infosreseaux@gmail.com</a><br>
> wrote:<br>
<br>
> Hi Daniel,<br>
><br>
> 2018-03-26 19:48 GMT+02:00 Daniel Patterson <<a href="mailto:daniel@mapbox.com" target="_blank" rel="noreferrer">daniel@mapbox.com</a>>:<br>
><br>
>> Hi François,<br>
>><br>
>> Yes, oneways onto your custom road network sound like exactly the kind<br>
>> of thing that the small component algorithm (<br>
>> <a href="https://en.wikipedia.org/wiki/Tarjan%27s_strongly_connected" rel="noreferrer noreferrer" target="_blank">https://en.wikipedia.org/wiki/Tarjan%27s_strongly_connected</a><br>
>> _components_algorithm) would flag. If your custom network is <1000<br>
>> nodes, it would've been flagged previously.<br>
>><br>
>> If you are only routing from the normal network *onto* your custom<br>
>> network, then changing the component size like you've done is the correct<br>
>> fix here. Just be aware, if you start on your custom network, and try to<br>
>> return to the main network, you will get "NoRoute" errors, but it sounds<br>
>> like that won't be a problem for you.<br>
>><br>
><br>
> This is indeed what I look for, no problem to get "No route" errors, but<br>
> only when network topology doesn't allow me to find one.<br>
> I don't understand why flagging part of the graph as "small element"<br>
> prevent a route to be found.<br>
><br>
> I would agree to a potential downgrade of small elements versus normal<br>
> ones, but a route have to be returned when it exists, haven't you?<br>
><br>
><br>
>> I'm not sure what's up with the turn icons in the debug viewer - make<br>
>> sure you've zoomed in far enough, turn icons only show up from Z12 and<br>
>> higher.<br>
>><br>
><br>
> Yes, I look between zoom 12 and 14<br>
> Still no turn icons<br>
><br>
> I'm wondering if it's only a display issue or a more serious problem in my<br>
> data processing.<br>
> Anyway, osrm gives me routes with turns and intersections so this sounds<br>
> to be only a display issue.<br>
><br>
><br>
> All the best<br>
><br>
> François<br>
><br>
> _______________________________________________<br>
> OSRM-talk mailing list<br>
> <a href="mailto:OSRM-talk@openstreetmap.org" target="_blank" rel="noreferrer">OSRM-talk@openstreetmap.org</a><br>
> <a href="https://lists.openstreetmap.org/listinfo/osrm-talk" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/osrm-talk</a><br>
><br>
><br>
-------------- next part --------------<br>
An HTML attachment was scrubbed...<br>
URL: <<a href="http://lists.openstreetmap.org/pipermail/osrm-talk/attachments/20180327/4c42bfb9/attachment-0001.html" rel="noreferrer noreferrer" target="_blank">http://lists.openstreetmap.org/pipermail/osrm-talk/attachments/20180327/4c42bfb9/attachment-0001.html</a>><br>
<br>
------------------------------<br>
<br>
Subject: Digest Footer<br>
<br>
_______________________________________________<br>
OSRM-talk mailing list<br>
<a href="mailto:OSRM-talk@openstreetmap.org" target="_blank" rel="noreferrer">OSRM-talk@openstreetmap.org</a><br>
<a href="https://lists.openstreetmap.org/listinfo/osrm-talk" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/osrm-talk</a><br>
<br>
<br>
------------------------------<br>
<br>
End of OSRM-talk Digest, Vol 63, Issue 11<br>
*****************************************<br>
<br>
_______________________________________________<br>
OSRM-talk mailing list<br>
<a href="mailto:OSRM-talk@openstreetmap.org" target="_blank" rel="noreferrer">OSRM-talk@openstreetmap.org</a><br>
<a href="https://lists.openstreetmap.org/listinfo/osrm-talk" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/osrm-talk</a><br>
</blockquote></div>