Skip to content


flagd architectures fall into two broad categories: those where the evaluation engine is deployed in a standalone process to which the client application connects (RPC), and those where the evaluation engine is embedded into the client application (in-process).

RPC vs In-Process Evaluation

RPC evaluation

In RPC-based deployments one or more flagd instances deployed and exposed to client applications in your infrastructure. flagd RPC providers use HTTP or gRPC to request flag evaluations from flagd. The request payload contains the flag key identifying the flag to be evaluated, as well as the relevant evaluation context. The flagd instance is configured to watch one or more syncs, and merges them to build its set of flags (see here for more details on flag definition merging). When sync sources are updated, flagd will send notifications to clients that flags have changed, enabling applications to react to changes by re-evaluating flags.

This architecture is can be leveraged by very simple clients, since no in-process engine is needed; in fact, you can evaluate flags directly from a terminal console using the cURL utility. One disadvantage of this pattern is the latency involved in the remote request (though flagd typically takes <10ms for an evaluation, and can evaluate thousands of flags per second).

title: RPC Evaluation
    flagd ||--o{ "sync (file)" : watches
    flagd ||--o{ "sync (http)" : polls
    flagd ||--o{ "sync (grpc)" : "sync.proto (gRPC/stream)"
    flagd ||--o{ "sync (kubernetes)" : watches
    "client app (+ flagd RPC provider)" ||--|| flagd : "evaluation.proto (gRPC/stream) / HTTP"

In-Process evaluation

In-process deployments embed the flagd evaluation engine directly into the client application through the use of an in-process provider. The in-process provider is connected via the sync protocol to an implementing gRPC service that provides the flag definitions. You can use flagd as a gRPC sync service. In this mode, the flag sync stream will expose aggregated flag configurations currently configured through syncs. This pattern requires an in-process implementation of the flagd evaluation engine but has the benefit of no I/O overhead for flag evaluations, since no inter-process communication is required.

title: In-Process Evaluation
    "client app (+ flagd in-process provider)" ||--|| "sync (grpc)" : "sync.proto (gRPC/stream)"