Analysys International CTO Guo Wei: Four ways to solve the cloud migration problem of Pb level big data cluster
This architecture can accelerate data parallel computing; Second, using the target analysis of the model and stratified sampling data, users can experience the results of millions of user actions in seconds, with an error less than.
second
Open source sample solution
Funnel Query
For many companies, the conversion rate from browser to product order payment and the loss rate of business development are necessary. Industry insiders call such a scenario "funnel query". In Guo Wei's opinion, the difficulty lies in "funnel query", and the orderly behavior order needs to be included in the "funnel". At present, most open source engines on the market are for unordered queries, which are difficult to query. In this regard, Yiguan provides on-site solutions. At present, the funnel of daily life is queried in seconds, and the source code will be opened in the near future.
summary
(Summary of big data migration)
Guo Wei believes that the migration of big data cloud includes several aspects: the basic framework of architecture, migration of historical data, parallel data verification, product switching and data management services. For several problems of capturing text, the international historical data of Yiguan has been switched seamlessly. At present, the big data analysis cloud platform has hundreds of millions of months of mixed living, about days of living, and hundreds of millions of cumulative installed capacity. It is also the first echelon in the Internet big data industry.
