Queries regarding Geode Architecture

[Goutham reddy <go...@gmail.com>]

Hi Team,
I am basically from Cassandra and NoSQL background. Can somebody help me
understand the below queries.

1. Considering a 5 node cluster, I believe servers are created on each
node(member), how many servers have to be created?

2. (5 node and 3 servers on each node) Creating a Replicate region does
that mean it is  replicated across all the servers and nodes?{5 nodes * 3
servers = 15 Regions}.

3. Is there a possibility to put two duplicate primary keys like Cassandra
(using clustering column).

4. We are implementing Microservices Architecture and we are considered
Redundant regions for two services. I read that Gemfire supports internal
messaging to update or delete other region. Can somebody explain how it
happens.

5. Last one, what is the exact difference between Peer to Peer topology and
Client Server Architecture.

Appreciate your help.

Thanks and Regards,
Goutham Reddy Aenugu.
-- 
Regards
Goutham Reddy

Re: Queries regarding Geode Architecture

[Anilkumar Gingade <ag...@pivotal.io>]

Goutham,

To be clear, when you refer "node" is it a physical box/machine or a JVM.
Geode allows you to configure Geode instance as a peer or Server (accepting
client connection). You can find more information on Geode Wiki page.

If you want to use GemFire (enterprise version of Geode); I would recommend
you to reach out to GemFire sales or customer support team; where you can
get professional help.

Regards,
-Anil.


On Tue, Mar 13, 2018 at 9:26 AM, Goutham reddy <go...@gmail.com>
wrote:

> Hi Team,
> I am basically from Cassandra and NoSQL background. Can somebody help me
> understand the below queries.
>
> 1. Considering a 5 node cluster, I believe servers are created on each
> node(member), how many servers have to be created?
>
> 2. (5 node and 3 servers on each node) Creating a Replicate region does
> that mean it is  replicated across all the servers and nodes?{5 nodes * 3
> servers = 15 Regions}.
>
> 3. Is there a possibility to put two duplicate primary keys like Cassandra
> (using clustering column).
>
> 4. We are implementing Microservices Architecture and we are considered
> Redundant regions for two services. I read that Gemfire supports internal
> messaging to update or delete other region. Can somebody explain how it
> happens.
>
> 5. Last one, what is the exact difference between Peer to Peer topology
> and Client Server Architecture.
>
> Appreciate your help.
>
> Thanks and Regards,
> Goutham Reddy Aenugu.
> --
> Regards
> Goutham Reddy
>

Re: Queries regarding Geode Architecture

[Anthony Baker <ab...@pivotal.io>]

Hi Goutham!

> On Mar 13, 2018, at 9:26 AM, Goutham reddy <go...@gmail.com> wrote:
> 
> Hi Team,
> I am basically from Cassandra and NoSQL background. Can somebody help me understand the below queries.
> 
> 1. Considering a 5 node cluster, I believe servers are created on each node(member), how many servers have to be created?
> 

Most deployments start one Geode process on each VM or physical host.  If you have five hosts, you could deploy a five member cluster.  Also note that you want to deploy 1+ Geode locators for discovery and management purposes.

Some deployments that focus on small heap sizes to avoid GC pauses will start more than process / host but that’s not as common.

> 2. (5 node and 3 servers on each node) Creating a Replicate region does that mean it is  replicated across all the servers and nodes?{5 nodes * 3 servers = 15 Regions}.

A replicate region will have the same copy of data on every member hosting that region.  Replicate regions are great for high read / low write use cases.  If you want to scale horizontally use partitioned regions.
> 
> 3. Is there a possibility to put two duplicate primary keys like Cassandra (using clustering column).

Keys are unique in Geode.  You can colocate related entities in separate partitioned regions using composite keys.

> 
> 4. We are implementing Microservices Architecture and we are considered Redundant regions for two services. I read that Gemfire supports internal messaging to update or delete other region. Can somebody explain how it happens.

Geode will automatically and synchronously update all copies of data.  If you have two separate entities that are updated together perhaps you should store them in the same region.  Geode can store and query nested objects.

> 
> 5. Last one, what is the exact difference between Peer to Peer topology and Client Server Architecture.
> 

A peer is a member of the cluster and typically hosts data.  A client connects to the cluster and may cache data locally for speed.  A typical microservice pattern is to create a SpringBoot application that accesses a Geode cluster using SpringDataGeode as a client.

> Appreciate your help.
> 
> Thanks and Regards,
> Goutham Reddy Aenugu.
> -- 
> Regards
> Goutham Reddy