Accumulo is designed to run on a large number of servers and includes several features designed to make administrating and maintaining large clusters tractable for administrators.
In particular, dynamic load balancing and automatic recovery from common types of hardware failure help keep Accumulo healthy even in clusters of over a thousand machines, in which hardware failures are common.
Here are the ways in which the software environment should be set up to support installing Accumulo. For suggestions on selecting appropriate types of hardware and sizing a cluster, see “Hardware Selection”.
Accumulo is regularly run and tested on several versions of Linux:
Red Hat Enterprise Linux
Ubuntu 12 and above
Development platforms include Linux and Mac OS X.
A few low-level kernel settings that can dramatically impact the responsiveness of the cluster might need to be tuned for scaling up the per-machine resources allowed when more than 100 machines are in a cluster.
Tablet servers should be given enough operational memory to avoid swapping. Swapping is bad because it can cause a tablet server to have to wait while the kernel retrieves from disk some page of memory that was swapped out. This delay can interfere with Accumulo’s ability to determine the responsiveness of the tablet servers and to keep all the data online.
To help avoid swapping, it is recommended that the Linux ...