High-Availability Evaluation

Part of the In-Memory Data Management Research book series (IMDM)


As shown in Fig.  8.3 in the previous chapter, the size of a RAMCloud Cluster can be changed without an interruption of the query processing executed by AnalyticsDB. However, the dynamic resizing in Fig.  8.3b is done via a purposeful revocation of a node which gives RAMCloud the time to redistribute the data from the to-be-removed node before its actual revocation. This kind of awareness cannot be expected in the event of a hardware failure. For the scenario of an unexpected hardware failure, RAMCloud features a fast crash recovery mechanism [ORS\({^+}\)11] as explained in Sect.  3.3.

Copyright information

© Springer International Publishing Switzerland 2016

Authors and Affiliations

  1. 1.Hasso Plattner InstituteUniversity of PotsdamPotsdamGermany

Personalised recommendations