Highlighted
DukeR1337
2 Iron

Issues with DDBoost 10 GbE and Data Domain with Veeam

Hi everyone

We have a physical backup server with 1 Gbit Ethernet port and 2x 10 Gbit Ethernet ports and two data domains systems in different locations with 1x 1 Gbit  and 2x 10 GbE ports for each system and DDBOOST licenced on both EMC Data Domains.

Each management and each 10 gig location is within an own VLAN and with own routers.

Backup Server

1 Gigabit

IP 192.168.1.10

Subnet 255.255.255.0

Gateway 192.168.1.1

10 Gigabit 1

IP 10.10.10.10

Subnet 255.255.255.0

GW N/A

10 Gigabit 2

IP 10.10.10.11

Subnet 255.255.255.0

GW N/A

Interface Group

Default

Int 10 Gig 1

Int 10 Gig 2

Allow all clients *

Allow all mtrees *

Static route 10.10.11.0 via gateway 10.10.0.1

Static route 10.10.12.0 via gateway 10.10.0.1

EMC Data Domain 1

1 Gigabit

IP 192.168.2.10

Subnet 255.255.255.0

Gateway 192.168.2.1

10 Gigabit 1

IP 10.10.11.10

Subnet 255.255.255.0

GW N/A


10 Gigabit 2

IP 10.10.11.11

Subnet 255.255.255.0

GW N/A

Interface Group

Default

Int 10 Gig 1

Int 10 Gig 2

Allow all clients *

Allow all mtrees *

Static route 10.10.10.0 via gateway 10.10.11.1

Static route 10.10.12.0 via gateway 10.10.11.1


EMC Data Domain 1

1 Gigabit

IP 192.168.3.10

Subnet 255.255.255.0

Gateway 192.168.2.1

10 Gigabit

IP 10.10.12.10

Subnet 255.255.255.0

GW N/A

IP 10.10.12.11

Subnet 255.255.255.0

GW N/A

Static route 10.10.10.0 via gateway 10.10.12.1

Static route 10.10.11.0 via gateway 10.10.12.1


EMC Datadomain

I've added the EMC Data Domain names with the management IP addresses on each EMC Data Domains

I've added the replication partrners with the names of each systems which points to the management (1gig) interfaces


Veeam

I've added the management ports of both EMC DDs into Veeam within Backup Repositories.


Communication with ping is working without any issues between each component and I am able to replicate from EMC 1 to EMC 2 and to backup data via Veeam to EMC DD1.


From Veeam to EMC Data Domain 1 I can see that the backup server and DD1 are using the 10 Gigs even I've specified the 1 Gig management ports which is exactly what I want and what is what I understand from DD Boost. They talk to each other and share there interface information and choose the best link in this case the 10 Gig.


BUT


From DD1 to DD2 it is not replicating via the 10 Gig ports but with the 1 Gig!!!! And this is what I don't understand. When DDboost is licensed and they exchange their information via management 1 gig ports it should detect the 10 Gig ports and replicate via that way but it does not!!! Why?


Do I have to add the systems in the host list with the 10 Gig IP addresses??? If so is it load sharing then ?? Do I have to set an additional static route the the host on each site???


Is there any logs, config page or commandlet to see why this happens and where to change or force the expecting behavior? Do I have issues understanding DDboost?


In other customer environments with less networks it works with the 1 Gig Management and 10 Gig replication links even I've added the systems with the 1 Gig management IP.......


Thanks for your help


PS: I don't use DNS Servers!!! I am using IPs from Veeam to EMC DD but the host list within EMC Data domain.

0 Kudos
4 Replies
umichklewis
3 Zinc

Re: Issues with DDBoost 10 GbE and Data Domain with Veeam

Quick question about DD replication - are you using array-based replication (mtree, collection, directory, etc.) or are you using file-managed replication in DDBoost?  If you're using array-based replication, you can modify the replication association to use the IP address of the 10GbE interfaces, instead of the 1GbE.

Let us know if that helps!

Karl

DukeR1337
2 Iron

Re: Issues with DDBoost 10 GbE and Data Domain with Veeam

Hi Karl

Thanks for your response. If I specify the IP address of the 10 Gig do I have to specify this in the host list and add the system with its name to the replication context?

If so, do I have to specify both 10 Gig IP addresses or only one? Is it able to load balance and failover then?

In addition do I missunderstand the function of DDboost with the 1 Gig IP address and 2x 10 Gig IP addresses? I always thought that I can setup the replication with the management IPs (1 Gig) and then specify the interface group for the replication.

Am I wrong? What is the reason that I have to do it your way?

Thank you very much Karl

PS: I found that one https://www.emc.com/collateral/TechnicalDocument/docu57271.pdf on page 17 there are some information can someone tell me if that is still up to date? Do i need to add the 1 Gig into the interface groups as well or has it to be seperate?

Note The managed file replication connection between the Data Domain systems is not part of IFGROUP. A single IP address is used for the destination Data Domain system. EMC recommends excluding one interface from the IFGROUP and reserving it for the managed file replication path between the source and destination Data Domain systems.

Also for array based??

0 Kudos
umichklewis
3 Zinc

Re: Issues with DDBoost 10 GbE and Data Domain with Veeam

With directory, mtree and collection replication, you can specify the target DNS name or IP in the GUI or CLI.  When creating or modifying the replication, go to the Create Pair window.  Go to the Advanced tab.  Specify the 10 Gig hostname/IP in the Connection Host field.  The Data Domain replication will now use this IP for the replication traffic:

2016-11-15 09_52_58-dd4200-prod-mgmt.bronsonhg.org - System Manager - Internet Explorer.png

In my case, I have a replication going from a DD4200 with 1Gb management and 10Gb management ports.  I specify the hostname/IP of the 10Gb interfaces in the above dialog box to use the 10Gb ports for replication traffic.

Let us know if that helps!

Karl

DukeR1337
2 Iron

Re: Issues with DDBoost 10 GbE and Data Domain with Veeam

So that is even another way, I was not aware of this but this means we can not use 2 Links for Failover or LB?

I am still confused because within IP Network we can specify the interface group for the mtree replication.

Shouldn't that work too?

Cheers

0 Kudos