Fast Cache, Fresh data. Can we have it all? Abstract Most national mapping authorities aim to provide a broad range of authoritative, up-to-date, easily accessable cached basemap services in accordance with industry standards. It is the adherence to these goals that set our cache services apart from the likes of Google and Bing, where a broad range of maps and update frequency have been neglected in order to acheive lightning fast, reliable, single services. Our responsibilities as a national provider of authoritative data require us to go further. The problem for national mapping authorities is that the consumers now demand the best of both worlds. When accessing our services, they not only assume an authoritatve date source that is correct at the time of viewing, but also demand performance on a par with that offered by the market leaders. It’s telling that both the private and public sectors find it extremely difficult to meet these expectiations; superfast, reliable cache services with data that is as close to real-time accurate as possible. This presentation will look at the Norwegian Mapping Authorities project to first find a solution to these probems, and then implement it through a completely opensource infrastructure. The project first came about as a result of increased user feedback concerning the cache services following the release of a new national mapping client, Norgeskart.no. We found that, primarily, the complaints were about the stability of the cache speed, not about the top speed itself. They encountered this problem because of the inherent issues in providing fresh data, namely an incomplete cache. The consumers expected our cache service quality indicators (speed, stability, uptime) to mirror those of the perceived commercial alternatives, and, due to various reasons, they were not. The initial primary goal of the project was ‘to increase the speed stability of our cache services, creating an infrastructure that enables them to run as close to top speed 100% of the time during normal conditions’. But through an analysis of the main issues, it was clear that to meet the primary goal, we would need to solve 2 other central problems; incremental updating of the caches and preseeding the caches through all the zoom levels. Our solution was actually implemented from 6 tasks, but these all stemmed from solving the 2 central problems. The upgraded infrastructure still uses many of the same components as the old (PostGIS, Mapserver, geowebcache, linux) but these are now utilised in a different way and the infrastrucure on the whole is configured very differently. We hope that our solution can act as an example for those national mapping authorities facing the same issues that we have faced.