cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ian Holsman <>
Subject Re: Is Cassandra suitable for multi criteria search engine
Date Fri, 18 Dec 2009 20:41:53 GMT
Hi David.
3 million is a good size. I would say it is a 'medium'
but it really depends on a lot of factors, and what exactly you are indexing.
as a rule of thumb if you can fit the index in memory you'll be fine. 
It also depends on how much of a long tail you have, how often you update the index (each
commit clears the caches)
and how complex your queries are. I've found the number of commits plays a bigger part than
the physical size.

You should get a full size index up and benchmark it, in normal operation to be sure.

you can also install Solr in 'distributed' mode, which lets you scale it out further.

On Dec 19, 2009, at 12:30 AM, David MARTIN wrote:

> Is a 3 million records set not a big deal for Solr? If I consider
> about 30 properties per item, I have to give Solr 90 millions
> properties to consider. Is that volume still correct for such a
> solution?
> And regarding lucene on top of Cassandra, can people share their feed
> back, if any, about such a solution. Pros & cons vs Solr for instance.
> Thank you.
> 2009/12/17, Jake Luciani <>:
>> True replication and scale.
>> On Dec 17, 2009, at 4:56 PM, Josh <> wrote:
>>> I've used solr a bunch (And I'd cosign gabriel: Solr's fantastic) and
>>> I'm trying to work my head around Cassandra, but I'm really hazy on
>>> what the Cassandra+Lucene combo gives you.  What are you trying to
>>> accomplish?  (Meant earnestly:  I'm really curious)
>>> josh
>>> @schulz
>>> On Thu, Dec 17, 2009 at 2:52 PM, Jake Luciani <>
>>> wrote:
>>>> You can also put lucene on top of Cassandra by using.
>>>> On Dec 17, 2009, at 4:43 PM, gabriele renzi <>
>>>> wrote:
>>>>> On Thu, Dec 17, 2009 at 7:48 PM, David MARTIN
>>>>> <>
>>>>> wrote:
>>>>>> Hi,
>>>>>> That's what I was thinking. And I'm glad to read Apache solr in
>>>>>> your
>>>>>> answer as it is one of my main leads.
>>>>> as a happy solr user, I second the suggestion, lucene (the
>>>>> technology
>>>>> behind solr) handles a number of documents like that without a
>>>>> sweat,
>>>>> and solr gives your replication and a few other good things.

Ian Holsman

View raw message