We have two different architectures. 1. If I need to send a mobile client a message I wait until the mobile client checks in e.g. the pattern is for the mobile client to ask and not to tell. 2. Some clients we run in a VPN so we essentially have a static ip provided by the Virtual IP. This uses a fair amount of bandwidth and the VPN gateway becomes a SPOF.
I would highly suggest going with option #1. Thrift does support SSL but some clients I don't think yet support Client and Server validation. In our case we built authentication that runs inside of SSL. -b On Tue, Mar 1, 2016 at 11:51 AM, Matt Chambers <[email protected]> wrote: > > > > > On Mar 1, 2016, at 11:44 AM, Thomas Witkowski < > [email protected]> wrote: > > > > 1. But if the back-end wants to make a remote procedure call to a mobile > device, so the back-end is the client and the mobile device is the server, > the mobile device must open the connection in some way before, as the > mobile device's IP is a dynamic IP. This is, what I thought by "register" > to the back-end. > > The mobile device would report into the server, but then the server has > connect back to the client. So you need thrift servers on both ends. > > > > > 4. You are right here. Does Thrift supports SSL to secure a service? > > > > Yes > > -- https://github.com/bearrito @deepbearrito
