Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (athena.apache.org: domain of iranitov@gmail.com designates
 209.85.212.169 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=message-id:date:from:user-agent:mime-version:to:subject:references
         :in-reply-to:content-type:content-transfer-encoding;
        b=WC3PjKwB4Uj41eaq6VvhOMQg7pOB2OzpYLwJc+pwwj4XS6LV3QD02Mzv/WM4wPuhIG
         lNdgjJnFp2rwzYSy4kw0/k+ZLa4Y466YWk1npxYKX2MW+TyuL6abEA5Ubqse9agXLxh+
         7y3Xkp9OjGkpBvGk1oWRCSmZtakzhYwQvd1Wk=
Message-ID: <4C02DD69.5010903@gmail.com>
Date: Sun, 30 May 2010 14:49:29 -0700
From: Igor Ranitovic <iranitov@gmail.com>
User-Agent: Thunderbird 2.0.0.23 (X11/20090817)
MIME-Version: 1.0
To: user@hbase.apache.org
Subject: Re: how to get the row before this one?
References: <3EF5E168-09A0-4FC9-ADBC-FC98D8172104@zerofootprint.net>
In-Reply-To: <3EF5E168-09A0-4FC9-ADBC-FC98D8172104@zerofootprint.net>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit

One thing that I can think of is to sort measurements from the newest to 
oldest:

DevA,12pm,150W
DevA,9am,100W

Now instead of scanning from-to date, start with the to date and keep 
scanning until you find a row with a date that is older than the from date.
To use your example, start the scan from 1pm and stop when you hit 9am.

To sort row from newest to oldest use MAXINT - timestamp, where 
timestamp is seconds since epoch.

Hope this helps.
i.

Oliver Meyn wrote:
> Hi all,
> 
> I have a table "device_power" which has as its row key 
> "device_id.timestamp", and a single column, "power".  A new row is 
> written whenever the power consumption of a device changes, which could 
> be every second, or could be as much as days.
> 
> When I do a query like "what was the average power of device X between 
> Date 1 and Date 2" I can get all keys where device_id matches X (using 
> PrefixFilter) and timestamp > Date1 and < Date2 (using either RowFilter 
> or setTimeRange).  But I also need the last row before my time range to 
> know what state the device was in at the start of my query period.  
> Maybe an example will help make it more obvious (my timestamps are long 
> integers, but this is easier to read):
> 
> DevA.9am, 100W
> DevB.10am, 200W
> DevB.11am, 150W
> DevA.12pm, 150W
> 
> And now the query "What was the average power of DevA btw 10am and 
> 1pm".  Is there a clever way to "get the first row before 10am that 
> matches PrefixFilter("DevA") ?
> 
> Thanks,
> Oliver
>