hadoop-hdfs-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sandy Ryza <sandy.r...@cloudera.com>
Subject Re: real time analytics on hadoop using spark or storm
Date Sat, 07 Dec 2013 01:45:44 GMT
As Azurry said, Spark Streaming can process data in small batches as well.
 An advantage of Spark Streaming over Storm is that the same code can be
used both for small and large batches.  Both Spark and Storm can be used
with Hadoop.

-Sandy


On Fri, Dec 6, 2013 at 5:27 PM, Azuryy Yu <azuryyyu@gmail.com> wrote:

> spark streaming is a mini job, which can update every 150ms, but storm is
> long live process.
>  On 2013-12-07 9:12 AM, "Jay Vyas" <jayunit100@gmail.com> wrote:
>
>> Spark increases performance by using distributed shared memory.
>>
>> Storm on the other hand gives you realtime performance by processing data
>> sets in small batches.
>>
>> The case for Spark is when you want a more sophisticated data processing.
>>
>> The case for Storm is when you have large volumes of incoming data and
>> you want to run a process every 1000 records.
>>
>> If you want a better comparison, try comparing spark-streaming with
>> storm.
>>
>>
>>
>>
>>
>>
>>
>> On Fri, Dec 6, 2013 at 7:04 PM, Smarty Juice <smarty.juice@gmail.com>wrote:
>>
>>> can anyone explain what is the clear difference between spark and storm
>>>
>>> what are the use case of storm and spark?
>>>
>>> can it be used without haddop?
>>>
>>> what are the pros and cons of running with or without hadoop?
>>>
>>> thanks
>>>
>>>
>>
>>
>> --
>> Jay Vyas
>> http://jayunit100.blogspot.com
>>
>

Mime
View raw message