After four exacting months, its done: MarkLogic and StorageReview worked together to develop the first actual NoSQL environment to benchmark storagethat clearly illustrates the benefits of flash coupled with the industry’s fastest interconnect. The dataset included the publicly- available Wikipedia XML collection because of the inclusion of multi-language and non-ASCII text. See which devices work best. The benchmark was run on a four node MarkLogic cluster driving extremely high I/O rates over Infiniband to the SSD device under test. Maximum ingestion rates were maintained to all four nodes in the face of continuous forest merge activity. Latencies were then measured to determine how the devices performed under this stressful load. The results? No real surprises …
(hat tip to MarkLogic’s Lead Performance Engineer Denis Sheahan for continuously re-writing scripts, changing up configurations and explaining all this to me.)
Diane Burley
Responsible for overall content strategy and developing integrated content delivery systems for MarkLogic. She is a former online executive with Gannett with astute business sense, a metaphorical communication style and no fear of technology. Diane has delivered speeches to global audiences on using technologies to transform business. She believes that regardless of industry or audience, "unless the content is highly relevant -- and perceived to be valuable by the individual or organization -- it is worthless."