Praktisches & Grundsätzliches zur Informatik


Von Clouds zur virtuellen (transparent verteilten) Cloud

In dem Maße, in dem sich cloud-basierte IT-Umgebungen den klassischen (in-house) IT-Umgebungen als überlegen erweisen, wächst die Notwendigkeit, die cloud-basierte Umgebung größerer Unternehmen auf mehr als nur einen Provider von Cloud-Services zu verteilen.

Diesen Bedarf gezielt zu unterstützen ist (wohl als erster) Basho angetreten. Mit Riak haben sie — interessanter Weise auf Open Source Basis — ein über zahlreiche Cloud-Storage-Provider verteiltes DBMS geschaffen.

Grob gesehen funktioniert es wie folgt (Zitat):

    In multi-datacenter replication, one cluster acts as a primary, or source, cluster. The primary cluster handles replication requests from one or more secondary, or sink, clusters (generally located in datacenters in other regions or countries). If the datacenter with the primary cluster goes down, a secondary cluster can take over as the primary cluster. In this sense, Riak's multi-datacenter capabilities are » masterless «.

    In multi-datacenter replication, there are two primary modes of operation: fullsync and real-time. In fullsync mode, a complete synchronization occurs between primary and secondary cluster(s). In real-time mode, continual, incremental synchronization occurs – replication is triggered by new updates. Fullsync is performed upon initial connection of a secondary cluster, and then periodic­ally (by default, every 360 minutes). Fullsync is also triggered if the TCP connection between primary and secondary cluster is severed and then recovered.

    Multiple concurrent TCP connections (approximately one per physical node) and processes are used to maximize performance and network utilization. This replication capability supports fullsync and real-time sync between clusters, as well as their simultaneous operation. By default, this connection is unidirectional, however, bidirectional replication can be achieved by establishing two unidirectional connections between clusters (where a cluster is both source and sink).

    Replication can be configured for all data in the cluster or on a per-bucket basis, which allows for replication of specific subsets of data.
Wissenswertes zu "Virtuelle Cloud" zusammengestellt durch Gebhard Greiter.
tags: seiteCloudUmgebungenTCP: Cloud1gegreit Umgebungen1gegreit TCP1gegreit