Hi, I have two Cisco 1760 routers in headquarteers which are connected via GRE/IPSec tunnels with branch offices routers (they are Cisco 1760 too)located in other cities. These two main office routers are connected to two different ISPs and GRE/IPSec VPNs connected to each brunch site router are configured on each main office router. This was made to provide fault tolerance. In normal conditions VPNs work on a single headquarteers router connected to the 1st ISP. If the link to 1st ISP becomes dead I direct VPN traffic to use the second headquarteeers router by changing static routes on both headquarteers router and brunch office routers. Everything was fine until the last few mounths when our ISP started to perform network equipment maintaince almost every week and it annoys us very much since it disrupts network connectivity. In this case I intended to use EIGRP as dynatic routing protocol instead of changing static routes in a case of link failure. I configured EIGRP on every router and in a case of link failure it works pretty well, I tested it. But implementing a dynamic routing on my network exerted an issue. I can pretty well ping hosts located in brunch offices's networks, but TCP sessions (SMTP, POP3, AD) drop unexpectedly. I presume that I did somethng wrong when configuring hub and spoke IPSec VPNs. Here are my questions 1. Should I use transport or tunnel IPSec mode when configuring GRE/IPSec. I tested both modes and it's seems to me there's no difference in this case. 2. Then configuring crypto map on a spoke router should I configure it like this crypto map Map1 10 ipsec-isakmp set peer x.x.x.x <- hub 1 set peer y.y.y.y <- hub 2 set transform-set cipher match address 110 or like this crypto map Map1 10 ipsec-isakmp set peer x.x.x.x set transform-set cipher match address 110 crypto map Map2 20 ipsec-isakmp set peer y.y.y.y set transform-set cipher match address 120 Thank you in for your answers in advance.