Good point. Basically if we use a single switch to connect 2 SRXs in a cluster we introduce the switch as a single point of failure here. If you are dead set on separating your cluster nodes with switches, use 2 separate switches, one for control, one for data and keep the traffic on different vlans.
Although technically this DOES work and is indeed supported, for all the reasons below, I would consider using this option carefully. HTH On 16 July 2012 11:20, Mike Devlin <[email protected]> wrote: > Although it can work, its recommended that you dont. > > Any latency spikes between the switches can cause clustering to split, and > you will suddenly be in a split brain scenario. > > I had a short talk with A-TAC about it a while back and they highly > recommended against it for our build out. > > > On Mon, Jul 16, 2012 at 5:16 AM, Mark Menzies <[email protected]> wrote: > >> Hiya bud >> >> Yes that can work here. >> >> Just make sure that the SRXs are less than 100ms apart and each sync >> connection, both fabric and control, is on separate VLANs. >> >> HTH >> >> >> >> On 16 July 2012 10:04, Spam <[email protected]> wrote: >> >> > Is it possible to connect 2 SRX devices together into a HA Cluster by >> > connecting >> > the Control & Fabric Interlinks via switches or must they be directly >> > connected. >> > >> > My planned setup is as follows: >> > >> > SRX<->Switch<->10GB Xconnect<->Switch<->SRX >> > >> > I can also give each connection is own dedicated VLAN if that would >> help. >> > >> > Spammy >> > >> > >> > _______________________________________________ >> > juniper-nsp mailing list [email protected] >> > https://puck.nether.net/mailman/listinfo/juniper-nsp >> > >> _______________________________________________ >> juniper-nsp mailing list [email protected] >> https://puck.nether.net/mailman/listinfo/juniper-nsp >> > > _______________________________________________ juniper-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/juniper-nsp
