Arandom stroll through Computer Science research, by Adrian Colyer

Experiences with approximating inquiries in Microsoft’s manufacturing big-data clusters Kandula et al., VLDB’19 I’ve been excited in regards to the possibility of approximate question processing in analytic groups for a few time, and also this paper defines its usage at scale in manufacturing. Microsoft’s data that are big have 10s of thousands of machines, and therefore are utilized by large number of … Continue reading Experiences with approximating inquiries in Microsoft’s manufacturing big-data groups

DDSketch: a quick and fully-mergeable quantile design with relative-error guarantees

DDSketch: an easy and fully-mergeable quantile sketch with relative-error guarantees Masson et al., VLDB’19 Datadog handles a lot of metrics – some clients have actually endpoints producing over 10M points per second! For reaction times (latencies) reporting an easy metric such as for example ‘average’ is close to worthless. Alternatively you want to understand what’s happening at various … Continue reading DDSketch: a quick and fully-mergeable quantile design with relative-error guarantees

SLOG: serializable, low-latency, geo-replicated deals

IPA: invariant-preserving applications for weakly constant replicated databases

IPA: invariant-preserving applications for weakly consistent replicated databases Balegas et al., VLDB’19 IPA for designers, pleased times! Continue reading →