Fast interactive work had a visible batch cost
Twenty-four controlled paired runs showed quotation-response latency improving under a revised resource allocation while the analytics batch took longer. We retained both results and the additional annual expense, rather than claim a free performance improvement.
