hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Vladimir Rodionov <vladrodio...@gmail.com>
Subject Re: Slow HBase write across data center
Date Thu, 29 Jun 2017 20:57:19 GMT
SND/RECEIVE buffer sizes (TCP/IP) must be increased from default 8KB for
WAN communications. That is TCP/IP feature

Google "TCP/IP send receive buffer size for WAN"

-Vlad

On Thu, Jun 29, 2017 at 11:27 AM, Steve Howard <stevedhoward@gmail.com>
wrote:

> 2 seconds for 100 puts that aggregate 50mb is 20 milliseconds per put.  Out
> of curiosity, how are you measuring the time taken for both HBase and the
> network latency?  How many sessions are doing the PUT's?  How far away is
> the other DC?  Are the HBase connections persistent, or are you doing the
> full connection setup for each PUT?
>
> What do you see if you write a simple socket server in the other DC and
> send it a 500kb payload (no HBase) in a loop of 100?
>
> On Thu, Jun 29, 2017 at 1:30 PM, Sreeram <sreeram.v@gmail.com> wrote:
>
> > Hi,
> >
> >  I am facing very slow write speed when I write to a HBase cluster in
> > a different data center. The network round trip takes 1 ms. Average
> > time taken for 100 PUTs (each 500KB) takes over 2 seconds. Any network
> > or OS parameter that I need to check ?
> >
> > Will appreciate any inputs from the community on this.
> >
> > Thanks,
> > Sreeram
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message