hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jean-Marc Spaggiari <jean-m...@spaggiari.org>
Subject Re: Limit number of columns in column family
Date Thu, 19 Sep 2013 06:23:05 GMT
Don't worry for the language ;)

I don't think there is any mecanism today to limit the number of columns
into a column family.

There might be multiple options but they will all have some drawback.

On option is to have a daily mapreduce job looking at each row and doing
the cleanup. This can work if you don't have millions of huge columns
because you will have to keep track of all of them to see how many you have
and how many you need to remove...

There might be some other options, like keep the index in the column name
so you know you need to remove all column with name < XXX where XXX is the
last index value minus the numbre of columns you can to keep.

etc.

JM


2013/9/18 M. BagherEsmaeily <mbesmaeily@gmail.com>

> any cell in the same row.
> Sorry because of my poor language!
>
>
> On Thu, Sep 19, 2013 at 9:28 AM, Jean-Marc Spaggiari <
> jean-marc@spaggiari.org> wrote:
>
> > Hi MBE,
> >
> > When you are saying "cells  with least timestamp being removed" you mean
> > versions of the same cell? Or any cell in the same row/cf?
> >
> > JM
> >
> >
> > 2013/9/18 M. BagherEsmaeily <mbesmaeily@gmail.com>
> >
> > > Hi,
> > > I have a column family that I want the number of columns on it has a
> > > specific limit, and when this number becomes greater than the limit,
> > cells
> > > with least timestamp being removed, like TTL on count not time.
> > > Please guide me to find best optimized way.
> > >
> > > Thanks.
> > > MBE
> > >
> >
>

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