Saturday 30 July 2016

DFS Concept

Namaskar,
Here you can find basic overview concept for DFS functional and technical. Will be updating a deep drive on this part as soon as possible. 


Distributed File System (DFS) is the standard data replication solution for DCN facilities used for BCP/DR requirements. The DFS of one location is configured to replicate data to the corresponding recovery location(s). In the event of a disaster or non-availability of the primary data repository, the service is re-directed to the recovery location and vice versa. Each Facility is to provide DFS services to all the clients operating in the shared network. DCN must follow Microsoft DFS based solution within the Microsoft Windows Server Operating System on dedicated server for all sites. The site with <500 users is to have DFS installed on the File Server in the centre. This document defines the standards to be followed when implementing DFS.

DFS SOLUTION OVERVIEW:
Shared space is to be created and allocated to a project/s on the storage area of the File server. The pointer of this shared drive/location is to be created on the DFS server. The DFS server of the specific location (primary) is to be configured for replication with the recovery location/s DFS server. The replications must be scheduled to take place at off working hours.

File Server
Repository were the project data lies
Primary Location
Location where DFS server data will be residing
Recovery Location
Location where DFS data will be replicated and used during DR
Hub
Primary location DFS server
Spoke
Recovery location DFS server

Prerequisites:
DFS servers must be hosted in Shared Server Farm VLAN.
 
Domain membership – All DFS servers must be member servers of the respective Regional Domain (DCN<Region>.com or GDN<Region>.com).
Domain Controller – Active Directory Schema must be extended by running the following command “adprep.exe /forestprep”. (Note: This commands needs to be executed only once in the forest.

DFS Logical Topology

 

  









Staging Folder Configurations Item
Configuration
Primary file server
Select required space for project data
Folder creation on File server
DFSR-Staging to be created in the above free space
Quota
Quota limit must be applied
Replication Group
Default staging path must be changed to the new path

Replication Configuration
DFS server to server replication is based on project specific replication – i.e. one server can replicate to multiple servers and also receive files from multiple locations.


Item
Configuration
Replication Group
Mention New replication group
Replication Group Type
Multipurpose replication group
Name and Domain
Assign the replication group created, and Data Distribution description about the replication group
Replication Group Members
Select at least two servers which will replicate each other
Topology
Hub & Spoke
Hub Members
Hub member – Primary (Location) server Spoke member – Recovery server
Replication Group Schedule and Bandwidth
Schedule the replication time. (Off Production Hours. Data must be replicated between DFS servers once per day.)
Primary Member
Mention the Server which will have the latest up-to-date information
Folder to Replicate
Assign the folders needed to be replicated
Local Path of the other server
Path of the other server folder to be replicated
 
  Folder Configurations for DR location
 

Item
Configuration
Source server
In replicated folder properties, select Only Share the folder
Publishing Method
Share the replicated folder
Permissions
Grant appropriate permissions

Post Replication Configurations
The following configuration items on the DFS management console ensure that the file servers will not consume large space for staging folders.


Item
Configuration
Space
Identify free space on Primary File Server
Folder & Quota
Creation of folder DFSR-Staging & apply hard Quota. Set quota to restrict beyond 5 GB
Staging Path
On respective replication group properties of DFS management in advance, change the default size to 1024 MB
      
Firewall rules to permit DFS replication traffic:
The below table outlines the TCP/UDP ports used by the traffic between Primary DFS to Recovery DFS servers that must be opened on the Internal Firewalls to permit DFS data replication traffic.




. Application Protocol
Protocol
Port
LDAP Server
UDP
389
NetBIOS Datagram Service
UDP
138
NetBIOS Session Service
TCP
139
Randomly allocated high TCP ports
TCP
RANDOM ports – 5000 to 5100
RPC
TCP
135
LDAP Server
TCP
389
SMB
TCP
445


All Firewall rules for the above mentioned TCP/UDP ports are to be bi-directional rules. Hard coding ports on the DFS server’s Registry for Replication:
1. Add the Internet key under: HKEY_LOCAL_MACHINE\Software\Microsoft\Rpc
2. Under the Internet key, add the values "Ports" (MULTI_SZ), "PortsInternetAvailable" (REG_SZ) and "UseInternetPorts" (REG_SZ).

The following patches must be installed to ensure smooth data replication across DFS servers.
KB905700, KB912154, KB912850, KB917953, KB920335, KB925377

BACKUP
The DFS servers must be backed up on a daily basis using the Facility Shared Backup service. Additionally backup the namespace of the DFS server using DFSutil (This utility can be installed from Windows Support Tools in Windows Server operating system CD). This will create a text file. Backup this file on daily basis as mentioned above. Perform this action before the daily backup.
 
RESTORATION:
In case of disaster restore the above created text file by using DFSutil.
Following are the useful additional information and best practices:
Number of DFS roots per server One, unless a hotfix is installed - Refer to this Microsoft KBarticle.
Number of links per DFS namespace:
>5,000 links for domain-based DFS
>50,000 links for stand-alone DFS
    • .Full Replication take place between the two servers only once in Hub & Spoke mode initially, henceforth only incremental replications take place between the two servers daily.
    • Bandwidth allocation – Plan with 256 kbps initially, observe the performance over a period of time and set the sizing based on real-time observation.
    • DFS Replication uses staging folders to act as caches for new and changed files to be replicated from sending members to receiving members. Also each replicated folder uses its own staging folder, and each staging folder has a configurable quota.
    • Staged files in the staging folders might consume a large portion of the volume and lead to disk full conditions.
    • Files are purged based on high and low watermarks, must be carefully set based on each replicated folder's replication activity and available disk space on the server.
    • Staging management - Configure the staging folder path to be the same for all replicated folders on a given volume and use a quota system to configure a quota over that folder. General example Drive name: \DFSR-Staging\Project folder name or Project Pseudo name. NOTE: This activity is to be carried out only after the initial replication gets completed - when a new replication group is configured.

    No comments:

    Post a Comment