This post is more than 5 years old
82 Posts
1
3188
Replication for File - VNX
Hello Friends,
I am working on File level replication (only NFS exports) from primary VNX to secondary VNX, Below are my questions
1. In storage pools for file, what is the difference between type "System and Mapped" pool ?
2.In our environment Primary VNX has 2 pools for file, SAS & SATA, both are of type "System". There are 7 File Sytems the has been created either on SAS or SATA pool. These are the file systems that need to be replicated to secondary VNX. In secondary VNX, already there is a storage pool for file and of type "Mapped pool" and enough free space is available to accommodate the file systems from primary VNX . Is it possible to replicate the file systems from primary to secondary VNX with different types of storage pools for file?
3. Set up a new Data Mover Interconnect : In the interfaces field where we need to select an IP address to make available to local side of the interconnect and IP addresss for Peer side of an interconnect, i believe both the IP's need to be in same VLAN, am i right? Is it also necessary those IP's need to be in same VLAN of other ineterfaces (for example the interface used for NFS exports)?
Thanks
Dino
umichklewis
1.2K Posts
1
November 3rd, 2013 01:00
You can find much of the information on the pools in the VNX Guide, Managing Volumes and File Systems with VNX AVM. This covers much of what you're looking for.
1. A System Pool is a pre-defined pool. It consists of disks that match one of the system-defined AVM (automatic volume management) pools. The system-defined pools do not normally have advanced data services, such as (auto-tiering or VNX compression) enabled.
A Mapped Pool is a dynamically-created pool. It consists of LUNs that have advanced data services (such as auto-tiering or VNX compression enabled.
2. Yes. You can replicate from one type of disk to another, the same way you can replicate from one pool to another. If you use the Unisphere Replication Wizard, it will list the available pools at the destination VNX. You can also specify the destination pool (or filesystem) from the CLI as well.
3. The Datamover Interconnect only requires the IPs to be in the same VLAN for replication purposes. For example, if the primary VNX has interfaces on VLANs 10, 20 and 30, the secondary VNX needs to have an interface on only one of those VLANs for replication to work. HOWEVER, if your clients expect to reach your secondary VNX on the same VLANs (IPs), you will need to provide the same VLAN to the secondary VNX. So, in your example, if your NFS exports are on VLAN A, your secondary VNX needs a port on VLAN A to offer them up to clients on the same IP addresses. However, if you intend to update DNS or you have static /etc/hosts files on your clients, you may not need to provide those VLANs.
At one customer site, the secondary VNX had half the VLANs the primary VNX had. However, in the event of a failover, the customer needed to use a VPN to access the DR site, which routed the clients to the right VLAN.
Let us know if that helps!
Karl
12122
82 Posts
0
November 4th, 2013 09:00
Thanks a ton Karl.
3. The Datamover Interconnect only requires the IPs to be in the same VLAN for replication purposes. For example, if the primary VNX has interfaces on VLANs 10, 20 and 30, the secondary VNX needs to have an interface on only one of those VLANs for replication to work. HOWEVER, if your clients expect to reach your secondary VNX on the same VLANs (IPs), you will need to provide the same VLAN to the secondary VNX. So, in your example, if your NFS exports are on VLAN A, your secondary VNX needs a port on VLAN A to offer them up to clients on the same IP addresses. However, if you intend to update DNS or you have static /etc/hosts files on your clients, you may not need to provide those VLANs.
From the above i have question to ask,
So does it mean "IP address" on local side of interconnect and "IP address" on peer side of interconnect need to be on same VLAN but can be on any VLAN?
For example : On the primary VNX we have interfaces on VLANs 46 and 47 and also secondary VNX has interfaces on VLANs 46 and 47, my question is related to replication session, can the Ip addresss on local side of interconnect and peer side of interconnect be on VLAN 100 or it has to be on either 46 or 47 for replication to take place?
Rainer_EMC
8.6K Posts
1
November 4th, 2013 14:00
There are two sides to this.
For just the pure data replication to work it is only necessary that:
- Control station on one side can talk via IP to the other control station for replication control
- Data mover on the one side can talk via IP to the other side data mover for data transfer
For this it doesn’t matter whether they are on the same VLAN or not.
Of course if also want to be able to failover then your DR system must be configured so that your clients are able to access it.
Whether that is on the same VLAN or not doesn’t matter.
IF you need that transparent (without having to reconfigure the clients) then you need extra care.
I would suggest for first take a look at the Replicator manual
12122
82 Posts
0
November 4th, 2013 21:00
Thanks Rainer_EMC
In our case, already 2 network interfaces have been created on secondary VNX with IP's on the same VLAN as of primary VNX to export NFS
I was asking for pure replication purpose... you explained it very clearly thanks
12122
82 Posts
0
November 12th, 2013 09:00
I have worked on File System Replication and everything looks good, now my question is, during replication is in progress, how NFS exports to be created & maintained on the secondary system? such that during a Failover or Switchover of a filesystem the clients can access the destination object immediately.
In simple how the NFS exports properties should be on secondary system while replication of file system is active??
Thanks in advance
Dino
Rainer_EMC
8.6K Posts
0
November 12th, 2013 09:00
If you use VDM based NFS exports they get replicated with the VDM
If not you have to do it yourself.
Keep in mind the regular Replicator failover will not transfer IP addresses – so might have to re-mount on the clients as well.
12122
82 Posts
0
November 12th, 2013 10:00
It is not a VDM based NFS exports,
So if i have to create the NFS exports on the secondary system, shall i give a dummy Path and rest all the fields (DATA MOVER, FILE SYSTEM, IP ADDRESSES OF READ/WRITE HOSTS) same as the primary one. So that during a switchover or failover just "Path" field alone can be modified. Will this approach on NFS export work??
Thanks
Rainer_EMC
8.6K Posts
0
November 12th, 2013 10:00
You don’t need dummy path – the file system are there on the secondary but readonly
I would create a script
You need to test it and work out the procedures.
Also keep in mind that the file handles most likely change so a client NFS remount is necessary.
If you need more you can engage your local resources or professional service for procedures and workaround.
12122
82 Posts
0
November 18th, 2013 09:00
Sorry for the late reply, first of all thanks.
Yes, i created a replication session for a file system followed with NFS export creation on the secondary side. Now, I tested by mounting the NFS export from the secondary side using one of its interface to a local server and i was able to see all the files been replicated and also it was a R/O file system.
Here is one more Question,
We know the Checkpoints will be created on both ends of the array on the pool we choose, my question is the checkpoint space occupied in the pool during replication session creation will remain constant thereafter or it will vary ?
dynamox
1 Rookie
1 Rookie
•
20.4K Posts
1
November 18th, 2013 21:00
it will very depending on change rate and size, as you know check points are used to provide point in time recovery.