This may be slightly off topic however I have a very unique situation where I need to provide two diverse paths to a major stock exchange. Each host may either use route A or B for any given reason to access this particular exchange using two distinct routers and target address. The applicatiOn running on these hosts must only see/use one target address this needs to be transparent as possible. NIC bonding/teaming on the host side isn't a viable solution because of the latency overhead same goes for vrrp/hsrp. I believe my only option here is to setup multiple default routes with a preferred path of some sort. This seems to be possible using ip route2 on Linux. This just seems wrong on many levels and I thought I would post here because I know there is something obvious I'm missing. Please clue me in. Thanks. Sent from my iPhone 3GS.
rodrick brown wrote:
This may be slightly off topic however I have a very unique situation where I need to provide two diverse paths to a major stock exchange. Each host may either use route A or B for any given reason to access this particular exchange using two distinct routers and target address.
Have you considered point-to-point circuits?
The applicatiOn running on these hosts must only see/use one target address this needs to be transparent as possible. NIC bonding/teaming on the host side isn't a viable solution because of the latency overhead same goes for vrrp/hsrp.
What latency do you mean when you talk about NIC bonding and VRRP? Peter
On Fri, 18 Dec 2009 19:46:42 EST, rodrick brown said:
The applicatiOn running on these hosts must only see/use one target address this needs to be transparent as possible. NIC bonding/teaming on the host side isn't a viable solution because of the latency overhead same goes for vrrp/hsrp.
If you're worried about the latency issues with NIC bonding, what do you intend to do about the speed-of-light latency from Chicago to NYC?
Anycast? http://www.nanog.org/meetings/nanog29/abstracts.php?pt=NjcxJm5hbm9nMjk=&nm=n anog29 Might need to know a little more about the layout here for a better answer. -Scott -----Original Message----- From: rodrick brown [mailto:rodrick.brown@gmail.com] Sent: Friday, December 18, 2009 7:47 PM To: nanog@nanog.org list Subject: Routing to multiple uplinks This may be slightly off topic however I have a very unique situation where I need to provide two diverse paths to a major stock exchange. Each host may either use route A or B for any given reason to access this particular exchange using two distinct routers and target address. The applicatiOn running on these hosts must only see/use one target address this needs to be transparent as possible. NIC bonding/teaming on the host side isn't a viable solution because of the latency overhead same goes for vrrp/hsrp. I believe my only option here is to setup multiple default routes with a preferred path of some sort. This seems to be possible using ip route2 on Linux. This just seems wrong on many levels and I thought I would post here because I know there is something obvious I'm missing. Please clue me in. Thanks. Sent from my iPhone 3GS.
Maybe I am missing something, but how does VRRP/HSRP cause latency? On 12/19/09 3:45 AM, Scott Berkman wrote:
Anycast? http://www.nanog.org/meetings/nanog29/abstracts.php?pt=NjcxJm5hbm9nMjk=&nm=n anog29
Might need to know a little more about the layout here for a better answer.
-Scott
-----Original Message----- From: rodrick brown [mailto:rodrick.brown@gmail.com] Sent: Friday, December 18, 2009 7:47 PM To: nanog@nanog.org list Subject: Routing to multiple uplinks
This may be slightly off topic however I have a very unique situation where I need to provide two diverse paths to a major stock exchange. Each host may either use route A or B for any given reason to access this particular exchange using two distinct routers and target address.
The applicatiOn running on these hosts must only see/use one target address this needs to be transparent as possible. NIC bonding/teaming on the host side isn't a viable solution because of the latency overhead same goes for vrrp/hsrp.
I believe my only option here is to setup multiple default routes with a preferred path of some sort. This seems to be possible using ip route2 on Linux.
This just seems wrong on many levels and I thought I would post here because I know there is something obvious I'm missing. Please clue me in.
Thanks.
Sent from my iPhone 3GS.
-- Steve King Network Engineer - Liquid Web, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
VRRP/HSRP does not cause latency the problem we faced prior was when links flapped or timed out this would be too much of a hindrance for our users to reconcile application state with various trading venues we are trading thousands upon thousands of trades a minute to various destinations. As stated before Path A and Path B are two distinct paths they do however provide identical services but application state is not preserved. A new session and state must be established if a user decides to switch between paths. Essentially we provide the ability for users either shutdown and start sending orders to Path A or Path B based on latency from our servers to these trading venues we're actively monitoring latency between both end points. The overall design is being driven by our rigorous application needs more than anything. The implementation is straight forward we receive a duplicate set of feeds from site A and site B and can also access various services coming from site A or site B however, at any given time a user will be sending/recieving data from one of those destinations. Never both simultaneously. So my question what is the best way to provide this type of redundancy at the host level? The application will only use one target address. On Sat, Dec 19, 2009 at 1:17 PM, Steven King <sking@kingrst.com> wrote:
Maybe I am missing something, but how does VRRP/HSRP cause latency?
On 12/19/09 3:45 AM, Scott Berkman wrote:
Anycast?
http://www.nanog.org/meetings/nanog29/abstracts.php?pt=NjcxJm5hbm9nMjk=&nm=n
anog29
Might need to know a little more about the layout here for a better answer.
-Scott
-----Original Message----- From: rodrick brown [mailto:rodrick.brown@gmail.com] Sent: Friday, December 18, 2009 7:47 PM To: nanog@nanog.org list Subject: Routing to multiple uplinks
This may be slightly off topic however I have a very unique situation where I need to provide two diverse paths to a major stock exchange. Each host may either use route A or B for any given reason to access this particular exchange using two distinct routers and target address.
The applicatiOn running on these hosts must only see/use one target address this needs to be transparent as possible. NIC bonding/teaming on the host side isn't a viable solution because of the latency overhead same goes for vrrp/hsrp.
I believe my only option here is to setup multiple default routes with a preferred path of some sort. This seems to be possible using ip route2 on Linux.
This just seems wrong on many levels and I thought I would post here because I know there is something obvious I'm missing. Please clue me in.
Thanks.
Sent from my iPhone 3GS.
-- Steve King
Network Engineer - Liquid Web, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
-- [ Rodrick R. Brown ] http://www.rodrickbrown.com http://www.linkedin.com/in/rodrickbrown
HSRP/VRRP can be tweaked to less than 1s fail over time. Can you provide a copy of your network map for analysis? GLBP might be a viable option as fail over is not actually an issue at that point. On 12/19/09 2:48 PM, Rodrick Brown wrote:
VRRP/HSRP does not cause latency the problem we faced prior was when links flapped or timed out this would be too much of a hindrance for our users to reconcile application state with various trading venues we are trading thousands upon thousands of trades a minute to various destinations.
As stated before Path A and Path B are two distinct paths they do however provide identical services but application state is not preserved. A new session and state must be established if a user decides to switch between paths.
Essentially we provide the ability for users either shutdown and start sending orders to Path A or Path B based on latency from our servers to these trading venues we're actively monitoring latency between both end points.
The overall design is being driven by our rigorous application needs more than anything.
The implementation is straight forward we receive a duplicate set of feeds from site A and site B and can also access various services coming from site A or site B however, at any given time a user will be sending/recieving data from one of those destinations. Never both simultaneously. So my question what is the best way to provide this type of redundancy at the host level?
The application will only use one target address.
On Sat, Dec 19, 2009 at 1:17 PM, Steven King <sking@kingrst.com <mailto:sking@kingrst.com>> wrote:
Maybe I am missing something, but how does VRRP/HSRP cause latency?
On 12/19/09 3:45 AM, Scott Berkman wrote: > Anycast? > http://www.nanog.org/meetings/nanog29/abstracts.php?pt=NjcxJm5hbm9nMjk=&nm=n <http://www.nanog.org/meetings/nanog29/abstracts.php?pt=NjcxJm5hbm9nMjk=&nm=n> > anog29 > > Might need to know a little more about the layout here for a better answer. > > -Scott > > -----Original Message----- > From: rodrick brown [mailto:rodrick.brown@gmail.com <mailto:rodrick.brown@gmail.com>] > Sent: Friday, December 18, 2009 7:47 PM > To: nanog@nanog.org <mailto:nanog@nanog.org> list > Subject: Routing to multiple uplinks > > This may be slightly off topic however I have a very unique situation > where I need to provide two diverse paths to a major stock exchange. > Each host may either use route A or B for any given reason to access > this particular exchange using two distinct routers and target address. > > The applicatiOn running on these hosts must only see/use one target > address this needs to be transparent as possible. NIC bonding/teaming > on the host side isn't a viable solution because of the latency > overhead same goes for vrrp/hsrp. > > I believe my only option here is to setup multiple default routes with > a preferred path of some sort. This seems to be possible using ip > route2 on Linux. > > This just seems wrong on many levels and I thought I would post here > because I know there is something obvious I'm missing. > Please clue me in. > > Thanks. > > Sent from my iPhone 3GS. > > > >
-- Steve King
Network Engineer - Liquid Web, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
-- [ Rodrick R. Brown ] http://www.rodrickbrown.com http://www.linkedin.com/in/rodrickbrown
-- Steve King Network Engineer - Liquid Web, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
The overall design is being driven by our rigorous application needs more than anything.
The implementation is straight forward we receive a duplicate set of feeds from site A and site B and can also access various services coming from site A or site B however, at any given time a user will be sending/recieving data from one of those destinations. Never both simultaneously. So my question what is the best way to provide this type of redundancy at the host level?
The application will only use one target address.
You've stated two seemingly contradictory things. 1) The User decides which paths to take yet the 2) application cannot see more than one path. First, this sounds like a User issue. The application with such rigorous requirements should have the features you need to manage this. Barring that... :) The mechanism the User uses to (manually) decide which path to take should make the election and handle the switchover in visibility. Presumably, since your application cannot tell when its switched destinations/paths it needs to be notified if the network makes a VRRP/HSRP decision. This all points to the mechanism you presumably already have in place for manual path decisions. If you are using your Linux box or whatever to make your path choices, simply have a script that sees if the path preferences have changed and use your method to notify the Application. If you are using a Cisco (or other) dedicated router, run something on the Application box or servers that will notice this change (if even by querying the router) so it can proactively detect this. You've asked for a technical suggestion but have not provided any detail about the actual constraints you have -- though you've implied them without context. Deepak Jain AiNET
Am I right in assuming that you're establishing application-layer sessions to two hosts with two different IP addresses (outside of your control) which provide (close to) identical services? If so, there's not much you can do outside of the application itself (at least if you want a semi-robust solution). You could try (reverse) load balancer, but even then the application session would have to be disconnected before being switched over to the other host.
As stated before Path A and Path B are two distinct paths they do however provide identical services but application state is not preserved. A new session and state must be established if a user decides to switch between paths.
Ivan Pepelnjak blog.ioshints.info / www.ioshints.info
participants (8)
-
Deepak Jain -
Ivan Pepelnjak -
Peter Hicks -
rodrick brown -
Rodrick Brown
-
Scott Berkman -
Steven King -
Valdis.Kletnieks＠vt.edu