We migrate DataSynapse grids without stopping the compute that sustains your business

Yes. The migration runs in waves with dual-run: the target platform operates in parallel with GridServer, and each service moves across once its result parity has been verified. The production grid never stops; the reference case is a tier-1 bank with zero downtime throughout the entire process.

It depends on the workload profile. Symphony is the natural target for low-latency Service-Oriented workloads (intraday risk, XVA, pricing); HTCondor removes the licence cost for high-throughput batch workloads; cloud and Kubernetes bring elasticity for closing peaks or full grid outsourcing. In practice many grids end up on a combination, and the initial service inventory is what decides the split.

With result-by-result validation during the dual-run: the same portfolios and scenarios run on both GridServer and the target platform, and the numerical outputs (VaR, sensitivities, actuarial reserves) are compared systematically before each service is signed off as migrated. Deviations are explained and documented — essential in front of internal audit and the supervisor.

Grid clients are not rewritten in one go. GridServer Driver APIs are mapped to their equivalents on the target platform and, where it helps, an adaptation layer is introduced so the computation engines (proprietary libraries, C++/Java/Python) keep working unchanged during the transition.