On Tue, Nov 27, 2= 012 at 4:45 PM, Jim Klucar <klucar@gmail.com> wrote:

Roshan,

Depending on what your cluste= r setup is and what the resolution of the time stamp is you could do someth= ing like this to spread the data around:

<timestamp-LSBs>-<string>-<reverse timestamp>

Using the LSBs of the timestamp as a uniform hash, then splitting on al= l possible hashes would spread things around a bit. If you do this, then al= l scans must check all hashes for data.

On Tue, Nov 27, 2012 at 1:25 PM, Keith T= urner <keith@deenlo.com> wrote:

On Tue, Nov 27, 2012 at 1:22 PM, Ro= shan Punnoose <roshanp@gmail.com> wrote:
Thanks!

The fact that you are using a binary tree behind= the scenes makes perfect sense. Btw, what do you use in the standalone (no= n native) implementation? Does it use a TreeMap?

When not using native code, ConcurrentSkipListMap is u= sed. =A0=A0
=A0

On Tue, Nov 27, 2012 at 12:57 PM, Keith = Turner <keith@deenlo.com> wrote:

On Tue, Nov 27, 2012 at 12:21 PM, R= oshan Punnoose <roshanp@gmail.com> wrote:
The <string> would most likely be a fixed set of strings that do not = change over time.

My question is if it is bad to use a r= everse index timestamp in the row id? Will it cause problems with the table= t splitting, compaction, and performance if the data is always being sent t= o the top of the tablet? If I define a split as everything prefixed with &l= t;string>, then the ingest will go to one tablet, but then I add a rever= se timestamp in the row, and that would mean I am always copying data to th= e top of the tablet. Will this cause performance issues? Or is it better to= append to a tablet?

I do not think it should matter. Ins= erts go into a C++ STL map on the tablet server if using the nativemap. =A0= I think the implementation of that is a balanced binary tree. =A0So I do n= ot think inserting at the beginning vs the end would make difference. =A0Th= at being said, I do not think I have tried this so I do not know if there w= ould be any suprises. =A0I would be interested in hearing about your experi= ences.

=A0

On Tue, Nov 2= 7, 2012 at 11:51 AM, Keith Turner <keith@deenlo.com> wrote:

Keith

On Tue= , Nov 27, 2012 at 10:41 AM, Roshan Punnoose <roshanp@gmail.com> wrote:

I want to have a table where the row will co= nsist of "<string>-<reverse index timestamp>". But th= is means that the data is always being prefixed to the beginning of the row= (or tablet if the row is large). Will this be a problem for compaction or = performance?=A0

Can you tell me more about what <string> is= ? =A0For example is it a hash or does it come from the set "foo1"= ,"foo2","foo3". =A0 How does it change over time? =A0I = think the answer to your question depends on what <string> is.

=A0

I don't know if I heard this correctly, but someone once= mentioned that making the row id the direct timestamp could cause performa= nce issues because data is always going to one tablet, but also because the= re is trouble splitting since it always appends to the tablet. Is this true= , is it similar to what could happen if I am always prefixing to a tablet?<= /div>

Yes using a timestamp for a row coul= d cause data from many clients to always go to the same tablet, which would= be bad for performance on a cluster.
=A0

Thanks!
Roshan<= /div>