Veeam B&R v7.0 Design Advice in a SAN replacement project

[massimiliano.rizzi]

Hello experts,

as part of a SAN replacement project for a new small business VMware customer, we are planning to replace a production 4 Gbit/s Fibre Channel HP 2012fc Dual Controller MSA with a new IBM Storwize V3700.

In addition to being out of warranty, the HP 2012fc Dual Controller MSA Storage is almost running out of disk space, so there will soon be not enough room for data growth.

We are planning to use Veeam B&R v7 and finally get rid of HP Data Protector Software.

Current production environment is relatively small and is described below:

==================================================
- 2x VMware ESX 3.5 hosts with plenty of CPU and memory resources
- paid vSphere license with a valid maintenance and support agreement in order to upgrade the ESX hosts
- less than 10 virtual machines (approximately 2 TB raw data size)
==================================================

In order to get the most out of Veeam B&R v7 using CBT for faster incremental backup, we are also planning to upgrade the ESX hosts to the latest supported version of ESXi and the virtual machines’ hardware to version 7 or later.

With regard to the disk storage for Veeam backups, multiple options will be available because the SAN replacement project will leave behind the following:

==================================================
- the old HP 2012fc Dual Controller MSA Storage
- a HP ProLiant ML350 G5 Server (not connected to the SAN) currently being used as the HP Data Protector media server
- a HP 20 MSA with a number of SATA 1.5Gb/s hard drives directly attached to the HP Data Protector media server above
- a Tape Autoloader directly attached to the HP Data Protector media server above
==================================================

It would be great if someone could kindly share his thoughts/recommendations with regard to designing the new Veeam B&R v7.0 Backup Infrastructure.

Thanks and Regards,

Massimiliano

[dellock6]

If your goal is to save on budget and reuse as much hardware as possible, I would first add Veeam as a VM with only server and enterprise manager roles on it (easier to protect it) and with also the proxy role, but disabled (you can activate it to do restores, its faster than a physical proxy bounded to network mode).
Then, I'll convert the G5 into the new physical Veeam proxy + repository, and connect to it in direct attach mode one of the two SAN that better fill your needs, in terms of both space and performances. Finally, I'll connect this server into the FC fabric and I'll use DirectSAN mode.

Luca.

[massimiliano.rizzi]

Hi Luca,

first of all thank you for taking the time to reply to my question (and best of luck in your new job BTW ).

Based on your reply, I understand that in this scenario direct SAN backups are the best way to go. I also understand that you recommend connecting the primary storage for Veeam backups in direct attach mode to the G5 (once it has been converted into the new physical Veeam proxy + repository) and the G5 itself into the FC fabric in order to use DirectSAN mode.

Since in this scenario the best/fastest storage for Veeam backups is definitely the old HP 2012fc Dual Controller MSA Storage, would you still recommend to direct attach it to the G5 (thus requiring additional HBAs inside the G5) or it will be fine to attach both the G5 and the 2012fc MSA Storage into the FC fabric and mask the LUNs for Veeam backups on the 2012fc MSA accordingly so that only the G5 can access them ?

Thank you again for your support.

Massimiliano

[dellock6]

Oh, this one too. I suggested Direct Attach because since you were swapping and replacing a SAN I wasn't sure you had free FC ports.
It can work in this way too. The advantage of DA is you do not consume fabric ports, and the whole connection bandwidth is available for it, but this is not a problem for those old SAN anyway, only really high end storage can saturate FC bandwidth, and only if it's something less than 8 Gb.

Luca.

[massimiliano.rizzi]

Again, thank you for taking the time to reply to my question.

I have one more (final) question, however.

Since all HP hardware is currently out of warranty and the customer seems to have extra budget for the new infrastructure, we are also thinking about the possibility to replace the two old HP hosts with a couple of brand new IBM System x3650 M4 hosts.

Each node canister of the IBM Storwize V3700 provides 4x 6 Gbps SAS ports so, if the customer can afford a couple of brand new hosts, I think the best choice for connecting them would be 6 Gbps SAS direct attach. As far as I know, some of the pros of shared SAS include the ease of setup and management and a cheaper price than 8Gb FC or 10Gb iSCSI/FCoE while distance and scalability are some of the cons. In this scenario all servers and storage will be in the same rack and while it may be tricky to add hosts in the future with shared SAS, the option to purchase a couple of additional SAS host interface cards for the IBM Storwize V3700 will allow us to scale up to six hosts having redundant paths. Plus, 8Gb FC seems a little overkill this time.

With regard to the disk storage for Veeam backups, in this scenario we will convert one of the old HP hosts (DL360 G5) into a new physical Veeam repository and directly attach to it the 2012fc MSA. With regard to the Veeam proxy, we will also add Veeam as a VM with only server and enterprise manager roles on it (like you mentioned above) in order to have a faster proxy than a physical proxy bound to network mode.

Finally, we could convert the remaining ML350 G5 Server into an additional physical Veeam proxy + repository, directly attach to it the HP 20 MSA Storage in order to ship Veeam backups to a remote site using the WAN accelerator.

Does it makes sense to you ?

Thanks again !

[dellock6]

Totally, nice design indeed

Luca.

[massimiliano.rizzi]

Glad to hear that !

Thank you very much for your support.

Massimiliano