cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sergey Murylev <sergeymury...@gmail.com>
Subject Re: regarding schema and suitability of cassandra
Date Fri, 11 Apr 2014 14:26:05 GMT
Actually if you want to use Cassandra you should store all user related
data in single row with user ID as primary key.

On 11/04/14 18:14, Prem Yadav wrote:
> Thanks. 
> For the use case, what should I be thinking about schema-wise. ?
>
> Thanks,
> Prem
>
>
> On Fri, Apr 11, 2014 at 2:16 PM, Sergey Murylev
> <sergeymurylev@gmail.com <mailto:sergeymurylev@gmail.com>> wrote:
>
>     Hi Prem,
>
>
>>     Also, I have heard that Cassandra doesn't perform will with high
>>     read ops. How true is that?
>     I think that it isn't true. Cassandra has very good read
>     performance. For more details you can look to benchmark
>     <http://planetcassandra.org/nosql-performance-benchmarks/#EndPoint>.
>
>>     How many read connections per machine can handle and how do I
>>     measure that in cassandra/
>     Cassandra uses one thread-per-client for remote procedure calls.
>     For a large number of client connections, this can cause excessive
>     memory usage for the thread stack. Connection pooling on the
>     client side is highly recommended.
>
>     --
>     Thanks,
>     Sergey
>
>
>     On 11/04/14 13:03, Prem Yadav wrote:
>>     Hi,
>>     I am now to cassandra and even though I am not familiar to the
>>     implementation and architecture of cassandra, Is struggle with
>>     how to best design the schema.
>>
>>     We have an application where we need to store huge amounts of
>>     data. Its a per user storage where we store a lot of data for
>>     each user and do a lot of random reads using userid.
>>     Initially, there will be a lot of writes and once it has
>>     stabilized, the reads will increase.
>>
>>     We are expecting to randomly read about 15 GB of data everyday.
>>     The reads will be per user id.
>>
>>     Could you please suggest an implementation and things I need to
>>     consider if I have to go with Cassandra. 
>>     Also, I have heard that Cassandra doesn't perform will with high
>>     read ops. How true is that? How many read connections per machine
>>     can handle and how do I measure that in cassandra/
>>
>>
>>     Thanks
>
>


Mime
View raw message