wbtiger
/
foundationdb
mirror of https://github.com/apple/foundationdb.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Improve three_data_hall / three_dc descriptions 

This patch includes descriptions of the transaction log configurations
for these redundancy modes, and corrects what I believe is an incorrect
statement about the minimum number of machines required to make
progress. It also clarifies that commits require synchronous
replication across datacenters.
This commit is contained in:
Adam Kocoloski 2018-12-14 16:29:44 -05:00
parent e18713450f
commit 0a6d423569
1 changed files with 2 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
documentation/sphinx/source/configuration.rst
View File

@ -355,10 +355,7 @@ FoundationDB will never use processes on the same machine for the replication of
FoundationDB replicates data to three machines, and at least three available machines are required to make progress. This is the recommended mode for a cluster of five or more machines in a single datacenter.
``three_data_hall`` mode
FoundationDB replicates data to three machines, and at least three available machines are required to make progress. Every piece of data that has been committed to storage servers
will be replicated onto three different data halls, and the cluster will
remain available after losing a single data hall and one machine in another
data hall.
FoundationDB stores data in triplicate, with one copy on a storage server in each of three data halls. The transaction logs are replicated four times, with two data halls containing two replicas apiece. Four available machines (two in each of two data halls) are therefore required to make progress. This configuration enables the cluster to remain available after losing a single data hall and one machine in another data hall.
Datacenter-aware mode
---------------------
@ -370,7 +367,7 @@ In addition to the more commonly used modes listed above, this version of Founda
``three_datacenter`` mode
*(for 5+ machines in 3 datacenters)*
FoundationDB attempts to replicate data across three datacenters and will stay up with only two available. Data is replicated 6 times. For maximum availability, you should use five coordination servers: two in two of the datacenters and one in the third datacenter.
FoundationDB attempts to replicate data across three datacenters and will stay up with only two available. Data is replicated 6 times. Transaction logs are stored in the same configuration as the ``three_data_hall`` mode, so commit latencies are tied to the latency between datacenters. For maximum availability, you should use five coordination servers: two in two of the datacenters and one in the third datacenter.
.. warning:: ``three_datacenter`` mode is not compatible with region configuration.