Hi,
At office, we have I ISPs.
I want to lightly monitor each link latency in order to decide several routing. For that, I have only one external server: 1 IP, it's an OVH dedicated server. The quick picture is http://s24.postimg.org/n3436z64l/defaul_route.png
Default route is via ISP1. If OVH-server pings IP1: - the request will go through ISP1: it's OK - the reply will go through ISP1: it's OK
If OVH-server pings IP2: - the request will go through ISP2: it's OK - the reply will go through ISP1: it's KO because the metric will involve ISP1
How could I tell the Office gateway (CentOS 6) "IF you reply to a ping request from the OVH server that came on IP2 THEN route it via IP2"
Am 15.11.2013 04:43, schrieb Mihamina RKTMB:
Hi,
At office, we have I ISPs.
I want to lightly monitor each link latency in order to decide several routing. For that, I have only one external server: 1 IP, it's an OVH dedicated server. The quick picture is http://s24.postimg.org/n3436z64l/defaul_route.png
Default route is via ISP1. If OVH-server pings IP1:
- the request will go through ISP1: it's OK
- the reply will go through ISP1: it's OK
If OVH-server pings IP2:
- the request will go through ISP2: it's OK
- the reply will go through ISP1: it's KO because the metric will
involve ISP1
How could I tell the Office gateway (CentOS 6) "IF you reply to a ping request from the OVH server that came on IP2 THEN route it via IP2"
You may study http://www.lartc.org/lartc.html to find out how to handle 2 uplinks.
Alexander
You should be careful of doing any kind of network latency sensitive work with providers such as OVH. The latency of their networks can be very unstable.
You can end up in a flip flopping state very easily.
On 15 November 2013 03:43, Mihamina RKTMB mihamina@rktmb.org wrote:
Hi,
At office, we have I ISPs.
I want to lightly monitor each link latency in order to decide several routing. For that, I have only one external server: 1 IP, it's an OVH dedicated server. The quick picture is http://s24.postimg.org/n3436z64l/defaul_route.png
Default route is via ISP1. If OVH-server pings IP1:
- the request will go through ISP1: it's OK
- the reply will go through ISP1: it's OK
If OVH-server pings IP2:
- the request will go through ISP2: it's OK
- the reply will go through ISP1: it's KO because the metric will
involve ISP1
How could I tell the Office gateway (CentOS 6) "IF you reply to a ping request from the OVH server that came on IP2 THEN route it via IP2"
-- +261 34 81 738 69
CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
On 11/15/2013 01:00 PM, Andrew Holway wrote:
You should be careful of doing any kind of network latency sensitive work with providers such as OVH. The latency of their networks can be very unstable.
You can end up in a flip flopping state very easily.
What is this OVH ? Can someone give me a link?
Eliezer
On Sat, Nov 16, 2013 at 4:49 PM, Eliezer Croitoru eliezer@ngtech.co.ilwrote:
On 11/15/2013 01:00 PM, Andrew Holway wrote:
You should be careful of doing any kind of network latency sensitive work with providers such as OVH. The latency of their networks can be very unstable.
You can end up in a flip flopping state very easily.
What is this OVH ? Can someone give me a link?
Eliezer _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
It is only a matter of metric latency from and to the data-center that matters.. There are many clients that will like their options. For clients around Europe and their globally fiber-optic channels it might worth. If you are from for example japan or AUS I am not that your clients RTT will be well tuned for usage with their infrastructure.
When and if you have a testing period of the service It is possible to ask for answers about the basic Network and service performance which seems reasonable to me as an ISP and as a DATACENTER. I would try to test them while paying them a bit and to see if they can commit them-self to your demands in a reasonable period of time.
Eliezer
On 11/16/2013 11:53 PM, Robin Polak wrote:
On Sat, Nov 16, 2013 at 4:49 PM, Eliezer Croitoru eliezer@ngtech.co.ilwrote:
On 11/15/2013 01:00 PM, Andrew Holway wrote:
You should be careful of doing any kind of network latency sensitive work with providers such as OVH. The latency of their networks can be very unstable.
You can end up in a flip flopping state very easily.
What is this OVH ? Can someone give me a link?
Eliezer _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Alexander Dalloz -
Andrew Holway -
Eliezer Croitoru -
Mihamina RKTMB -
Robin Polak