Getting Started with CIFS Scale-out Cluster
Synology CIFS Scale-out Cluster allows you to create a clustered storage unit by aggregating the storage and computational resources of multiple Synology NAS servers. A CIFS Scale-out Cluster implements scaling-on-demand architecture, allowing storage capacity to be expanded linearly when adding more servers to the cluster.
In a cluster configuration, Synology NAS servers can be designated to assume the following roles -- Storage Server or Computing Server. Storage servers contribute hard drives to the cluster storage space. Computing servers handle and distribute file access requests, thus increasing the data throughput of the cluster. The same server can be designated as a computing server and storage server at the same time.
Term Explanation
- Cluster Volume: A cluster volume is an aggregated data storage pool consisting of 1 to 12 Synology NAS servers. Each server contributes specified hard drives, which are configured in a RAID 1/5/6 layout and constitute the storage units of the cluster volume. Only unused hard drives can be used to create a cluster volume.
- Storage Unit: A storage unit is composed of hard drives configured in RAID/1/5/6 on a storage server. Storage units constitute a cluster volume.
- Cluster Share: A cluster share is a shared folder that is created on a cluster volume in order to organize and share data. A cluster share can be accessed by users via CIFS protocol.
- Storage Server: A storage server is a Synology NAS server that contributes specified hard drives to be a part of the storage space of a cluster volume.
- Computing Server: A computing server is a Synology NAS server that provides computation power to handle and distribute file access requests to multiple storage servers. It also provides CIFS service, allowing users or applications to connect and access data. You can designate 1 to 12 servers to act as computing servers. The IOPS (Input/Output Operations Per Second) of the cluster increases linearly according to the number of designated computing servers.
Before You Start
Please see the following notes:
- Only certain Synology NAS models support the CIFS Scale-out Cluster feature.
- All servers to be included in the cluster must be joined to the CMS Host at CMS > Server > Add.
Setting up CIFS Scale-out Cluster
Once all of the servers have been joined to the CMS Host, you can follow the instructions below to set up the CIFS Scale-out Cluster.
To set up a new cluster:
- Go to CMS > Cluster Overview and click Cluster Settings.
- A setup wizard appears, prompting you to complete the following steps:
- Create Cluster Volume: Create a cluster volume on the storage servers for storing data.
- Select Computing Server: Computing servers are used for processing file access requests issued by clients. All servers must belong to the same domain/LDAP, so you need to enter the address and login credentials of your domain/LDAP server. If you don't have a domain/LDAP server, you can download and install Synology's Directory Server package in order to set up your own LDAP server.
- Setup Clustered CIFS Service: You can access the clustered CIFS service via a unified NetBIOS name or CIFS server IP address of your choosing.
- Create Cluster Share: Create a shared folder on top of cluster volume for organizing and sharing data.
Computing Server Load Balancing
After your cluster is running, we recommend setting up load balancing for the computing servers. Because all computing servers handle CIFS service concurrently, load balancing is necessary in order to distribute file access requests across all computing servers. There are two types of load balancing -- Load Balancing by NetBIOS Name and Load Balancing by Round Robin DNS.
Load Balancing by NetBIOS Name
With this method, when a client uses the NetBIOS name to access a cluster share, one of the computing servers responds to the request randomly, thus balancing the connection load among all computing servers.
Load Balancing by Round Robin DNS
Round robin DNS distributes the connection load of computing servers by directing queries to computing servers according to a rotating sequence. With round robin DNS, each query is directed to a different computing server, thus distributing the overall connection load among servers.
Round robin can be configured on the DNS server in your environment. If you don't have a DNS server in your environment, you can install Synology's DSN Server package and configure your DNS resource record.
To set up round robin DNS:
- Install and set up DNS Server according to this tutorial.
- Go to the Zones page, select a master zone, and click Edit > Resource Record.
- Click Create > A Type. Enter a hostname (e.g. if you cluster name and FQDN is [Your Hostname].[Your Domain Name], then the hostname is cluster.synology.com) and enter the IP address of one of your computing servers.
- Repeat the step above until all computing server are included in the A record. Please note, the hostname must be identical for every A record for computing servers.
- Point your computer to the newly configured DNS server. You should be able to access the clustered CIFS server by using the FQDN (e.g. cluster.synology.com).