How is failover expected to work in websockets api. In RFA I can have multiple ads hosts configured, comma separated , and the failover is taken care of, but ic an only supply one host in websockets.
How is failover expected to work in websockets api. In RFA I can have multiple ads hosts configured, comma separated , and the failover is taken care of, but ic an only supply one host in websockets.
Hi @ranjit.talukdar,
RFA is client-side compiled API provided by Thomson Reuters that offers a rich service to application developers. However, application developers are limited in terms of programming languages and hardware they can develop and run their applications. The WebSocket API is a server side protocol, not a client-side API. As such, if a client wishes to use this service, they are responsible for building failover and recovery logic. The upside of using the WebSocket service is the ability to utilize any technology on any hardware platform that supports WebSockets. A server side API that supports modern technologies, gives the application developer unlimited choices in terms of language and hardware platforms to run their applications.
thanks @nick.zincone.1
would you be able to give a suggestion on how to manage it client side. Would using the Ping:Pong messages be a solution?
cheers
Ranjit
Hi @ranjit.talukdar,
There are a few options you can use when working with fail over. I would recommend to simply use the connection status. For example, if a connection fails or goes down, walk through the list of servers that you define and attempt connect until success. In addition to this, you can also use the built-in PING messages sent from the ADS. If you have not seen an expected PING for a preset time, then disconnect from that server and walk down the line of your list of servers and try them.
Hi Nick,
May I ask for the definition of a server side and client side API please ?
We use websockets to consume data as a consumer. We connect to the ADS in a similar way to normal SSL and RSSL connections,
I am surprised there is no mechanism for failover
regards
Graham