Thank you Anoop. Though it's a bit strange to include CF in the index, since all the block index is contained in a HFile for a specific CF, I'm sure there would be a good reason (maybe for the performance of the comparison). Anyways it should be almost no issue since the length of the CF should be short(mostly one byte).
Thanks. -----Original Message----- From: Anoop John [mailto:[email protected]] Sent: Thursday, August 07, 2014 1:40 PM To: [email protected] Subject: Re: What is in a HBase block index entry? It will be the key of the KeyValue. Key includes rk + cf + qualifier + ts + type. So all these part of key. Your annswer#1 is correct (but with addition of type also).. Hope this make it clear for you. -Anoop- On Tue, Aug 5, 2014 at 9:43 AM, innowireless TaeYun Kim < [email protected]> wrote: > So, is it safe to assume that there is no documentation for the exact > content of the block index? > I think that reading the source code should be the last resort, since > one cannot sure whether it is an implementation detail, or it is the > specification that can be relied upon. > The information on the exact content of the block index is important, > since it is related in the size of the index (let alone the query > performance) and therefore the schema design. > > Or, could you please simply provide the exact answer to my original > question(with proper reference)? > > Thank you. > > -----Original Message----- > From: Ted Yu [mailto:[email protected]] > Sent: Tuesday, August 05, 2014 12:59 PM > To: [email protected] > Subject: Re: What is in a HBase block index entry? > > I suggest you read the source code of KeyValue class. > e.g. you can start with this method: > > public static long getKeyDataStructureSize(int rlength, int flength, > int > qlength) { > > Cheers > > > On Mon, Aug 4, 2014 at 8:13 PM, innowireless TaeYun Kim < > [email protected]> wrote: > > > So, your answer is 1. in my list? > > I don’t think so since the column family information is not necessary. > > Could you please refer a more direct information? > > > > > > -----Original Message----- > > From: Ted Yu [mailto:[email protected]] > > Sent: Tuesday, August 05, 2014 12:01 PM > > To: [email protected] > > Subject: Re: What is in a HBase block index entry? > > > > Please see: > > http://hbase.apache.org/book.html#keyvalue > > > > > > On Mon, Aug 4, 2014 at 7:55 PM, innowireless TaeYun Kim < > > [email protected]> wrote: > > > > > Thank you for your reply. > > > > > > It only says 'Key'. > > > That's what I'm confused. > > > > > > > > > -----Original Message----- > > > From: Ted Yu [mailto:[email protected]] > > > Sent: Tuesday, August 05, 2014 11:45 AM > > > To: [email protected] > > > Subject: Re: What is in a HBase block index entry? > > > > > > Have you read this ? > > > http://hbase.apache.org/book.html#d3593e20175 > > > > > > Cheers > > > > > > > > > On Mon, Aug 4, 2014 at 7:24 PM, innowireless TaeYun Kim < > > > [email protected]> wrote: > > > > > > > Hi, > > > > > > > > This is a newbie question. > > > > > > > > > > > > > > > > What is in a HBase block index entry? > > > > > > > > > > > > > > > > My guess is that it's one of these: > > > > > > > > > > > > > > > > 1. all key components: rowkey + column family + column > qualifier + > > > > timestamp > > > > > > > > 2. all key components except for column family (since the index > is > > > in > > > > a > > > > HFile that is a part of the storage for a column family): rowkey > > > > + column qualifier + timestamp > > > > > > > > 3. rowkey only > > > > > > > > > > > > > > > > I've tried to find the information, But the articles only say > > > > that it contains a 'key'. For me as a newbie, it is confusing > > > > since in a KeyValue the all key components comprise the 'key', > > > > while a rowkey is > > > also a 'key'. > > > > > > > > > > > > > > > > Thanks. > > > > > > > > > > > > > > > > > > > > > > > > > > > >
