====== Sharding configuration and QoE cluster setup ======
{{indexmenu_n>7}}

QoE Stor can operate in sharding mode — horizontal cluster scaling, where parts of a single database are distributed across different shards (nodes).

Sharding allows you to:
  - Distribute the IPFIX insertion load from multiple DPI units by spreading it across nodes.
  - Significantly accelerate report generation by leveraging the combined computational power of all nodes — reports are built in parallel and merged into a final result.

===== 1. Cluster creation =====

To create a QoE Stor cluster, configure the cluster node settings on each node in the file ''/etc/clickhouse-server/config.d/qoestor-cluster-config.xml''.

Edit the file on the first node:

<code>
<?xml version="1.0"?>
<yandex>
    <remote_servers>
        <qoestor_cluster>
            <shard>
                <replica>
                    <host>127.0.0.1</host>
                    <port>9000</port>
                </replica>
            </shard>
            <shard>
                <replica>
                    <host>127.0.0.2</host>
                    <port>9000</port>
                </replica>
            </shard>
            <shard>
                <replica>
                    <host>127.0.0.3</host>
                    <port>9000</port>
                </replica>
            </shard>
        </qoestor_cluster>
    </remote_servers>
</yandex>
</code>

Specify the IP address or hostname for each shard. Save the changes.

Copy the file to the remaining cluster nodes. 

Restart the database with the ''fastor-db-restart'' command on each node. The cluster is now ready for operation.

<note important>Port 9000 must be open on all cluster nodes in the internal network between nodes.</note>


===== 2. Adding the cluster in GUI =====

==== 1. Master node connection ====

Select which node will act as the master. Connect it in the GUI settings and enable cluster mode. See the screenshot below.

{{ :dpi:qoe_analytics:implementation_administration:configuration_setup:sharding:sharding-master-node.png?&1100 |}}

<note important>Port 8123 must be open on all cluster nodes.</note>

==== 2. Management via SSH ====

In the GUI section **Administrator / Equipment**, add all nodes for SSH management. 

{{ :dpi:qoe_analytics:implementation_administration:configuration_setup:sharding:sharding-ssh.png?&1100 |}}


==== 3. IPFIX receiver configuration ====

In the QoE Stor configuration, set up receivers for each node. 

{{ :dpi:qoe_analytics:implementation_administration:configuration_setup:sharding:sharding-receivers.png?&1100 |}}

Reconfigure the DPI to send IPFIX to the corresponding QoE cluster nodes.