I understand that part. But how do you prevent people starting a
rogue node and adding it to the system? As I understand it now,
anybody can bring up a node, point it at one of the seeds and have it
take part in the cluster. Am I mistaken there?
---Mark
On Fri, Aug 21, 2009 at 10:42 PM, Chris Goffinet<[email protected]> wrote:
> Thrift is just a cross-platform interface. Using the internal api does not
> mitigate having Cassandra find other nodes.
>
> -Chris
>
> On Aug 21, 2009, at 10:39 PM, Mark McBride wrote:
>
>> There's still the question of inter-node communication though. One of
>> the attractive things to us is the ability to power on another virtual
>> appliance and have it auto-discover the other Cassandra nodes. Is
>> this just something outside the scope of the current design?
>>
>> ---Mark
>>
>> On Fri, Aug 21, 2009 at 10:30 PM, Jonathan Ellis<[email protected]> wrote:
>>>
>>> if your product is jvm based, just use the internal api and don't
>>> stzrt the thrift listeners at all.
>>>
>>> On 8/21/09, Mark McBride <[email protected]> wrote:
>>>>
>>>> I'm looking at the potential of embedding Cassandra in one of our
>>>> products. This ships as one or more virtual appliances that runs at a
>>>> customer's site, and security is always an issue. This looks like
>>>> mostly a Thrift issue... but I was wondering if anybody on this list
>>>> had any thoughts about how you would go about securing Cassandra. The
>>>> best idea I have so far is to try to get THttpClient working (doc
>>>> there is very sparse), have Cassandra listen only listen on 127.0.0.1
>>>> and have Apache + mod_proxy handle security. If anybody thinks this
>>>> is a dumb way to do it I'm more than willing to listen to alternatives
>>>>
>>>> ---Mark
>>>>
>>>
>
>
