accumulo-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eric Newton <>
Subject Re: Robustness of MemTable
Date Mon, 30 Jul 2012 19:33:19 GMT
Yes, the key/value pairs are replicated in redundant logs on other machines.

We run long tests in which we ingest data and kill random servers;
this mechanism is robust up to the point to which it is configured.
That is, if you lose the right number of machines, you will not be
able to recover the in-memory data.  The default number of node
failures needed to prevent recovery is 2, though this can be changed
by configuration.


On Mon, Jul 30, 2012 at 3:19 PM, Edmon Begoli <> wrote:
> What is the redundancy/recovery mechanism of Write-Ahead Log and MemTable.
> I am specifically asking about this statement and the parts about:
> "When a write arrives at a TabletServer it is written to a Write-Ahead
> Log and then inserted into a sorted data structure in memory called a
> MemTable.
> ...
> If a TabletServer fails, the Master detects it and automatically
> reassigns the tablets assigned from the failed server to other
> servers.
> Any key-value pairs that were in memory at the time the TabletServer
> are automatically reapplied from the Write-Ahead Log to prevent any
> loss of data."
> I.e. what happens if machine holding a MemTable crashes. Will
> key-value pairs that we in memory have been replicated already?
> Thank you,
> Edmon

View raw message