Skip to main content

I have a customer looking to deploy Veeam but I’m trying to work out the pros and cons if we go with one of two options….

  1. Data Domain
  2. NetApp FAS storage fronted by a server and acting as a hardened repo

Would be good to get real world experiences, pros and cons etc.

Workloads are very traditional, VMs, Agents, SQL, Oracle, hyper-v, SAP, NAS and exchange

thanks

Mi Mike,

 

i work at a netapp only partner, so i deal with Ontap/FAS very often.

My opinion:

Never use Data Domain as primary Backup.

Ontap Storage/FAS is Nice - but not ideal for hardened Repo.

 

Better choose eSeries from Netapp as simple but fast Block storage. Put a simple Rack Server in front of this.

I posted something about this:

 

 

Regards

Matze


I would go for the Netapp. Worked with both systems. DataDomain is good at dedupe, but performance is absent. Especially for primary location I would go for a system with higher IOPS. I’ve tested SureBackup on this and had to tweak timeouts to get it running on DataDomain...


Never used Netapp, but I am with the guys. Do not use Data Domain as primary repo because the performance for restore is bad because of the dedup. If you want to use he is great for a secondary copy.


I am will add to the NetApp option.  I have never used them myself but have Data Domain.  They are fantastic for space savings but that is about it as restore performance is lacking especially for large system restores.


I agree with the opinions of ​@MatzeB and ​@kristofpoppe , we are also partners of both brands and do the same configurations on our customers.
There are no problems with any backup jobs.


Mi Mike,

 

i work at a netapp only partner, so i deal with Ontap/FAS very often.

My opinion:

Never use Data Domain as primary Backup.

Ontap Storage/FAS is Nice - but not ideal for hardened Repo.

 

Better choose eSeries from Netapp as simple but fast Block storage. Put a simple Rack Server in front of this.

I posted something about this:

 

 

Regards

Matze

great thanks very much. Excellent detail and will read the PDF as well. Appreciate it


TODAY new E-Series E4000 was launched - “old” E2800 is EOA soon ;) It’s very fast - already tested it ;)


Agreed with all above, the key point is that DataDomains are valid as secondary location but never as primary. Backup performance is...okay, but restore performance is extremly slow!


TODAY new E-Series E4000 was launched - “old” E2800 is EOA soon ;) It’s very fast - already tested it ;)

thanks Mattias, when did you get your hands on the E4000? I dont think its released in the UK yet. any links to the product datasheet? What the enhancements are etc….. thank you


Agree with all mentioned posts (never ever use a DD as a primary target…) but I want to also add a point regards the long term retention possibilities with a DD: 

  • with an enabled retention lock compliance mode on the DD, you have a very good immutability function (compliance is the only supported mode within VBR). It’s not even possible to overwrite it with admin/root privileges.

 

compared to the costs and if long term retention/dedup isn’t a topic, maybe a NetApp repository could be a better/cheaper solution. But here I would go with the expertise from Matze and colleagues.

 

edit: not NetApp related, but if the amount of data and other requirements fits into it, we have very positive feedback with some Dell R760xd2 systems, direct installed with VHR. 

 

 


TODAY new E-Series E4000 was launched - “old” E2800 is EOA soon ;) It’s very fast - already tested it ;)

thanks Mattias, when did you get your hands on the E4000? I dont think its released in the UK yet. any links to the product datasheet? What the enhancements are etc….. thank you

I got a E4000 some months ago, but as an NDA sample for benchmarks. The Netapp and Veeam Eseries TR will be upgraded with this Informations in the next time.

 

Officially the System was launched today. It's Quatable, Specs in HWU etc. So you should be able to find All Informations easily. 

The Performance is really great for an HDD System, I meassured around 5-6GB/s as hardened Repo with 60 HDDs.

If you need details send me a pm. 

 

Regards 

Matze


My advice is to look at the entire architecture and design a plan before picking specific storage.

Step 1 - Find out how much space you need.

This is the most critical step. Combine it with how many sites you plan on using as well. Is the second location going to be another datacenter or cloud?  

Many places have different policies such as 30 days at the main site, and then GFS at a secondary site. If you go Immutable, (which everyone should) keep in mind the additional restore points that may be kept to remain in policy. 

After using the Veeam Calculators, make sure to add some room for growth, and keep in mind an 80% max is usually recommended for storage due to overhead, metatdata, swing space etc. 

 

Step 2 - What are your performance requirements.

As stated above, dedupe shouldn’t be your primary backup, and can take a long time to restore. If you plan on doing instant restores, sure backup in large scale, and other tasks that use lots of IO, there may be performance issues. Dedupe appliances are great for saving space and having long term retention. This is a great secondary option. 

I have NVMe storage at my primary site and you can even put that in a SOBR with some slower disk if you size your pools correctly. This is great when I want to provide SureBackup labs for our apps teams to do testing of updates on their systems. I have slower tired disk at another site on immutable storage holding GFS copies for long term retention.

Step 3- Figure out your budget.

It’s easy to get swept away by a good sales pitch. Make sure you know your budget so you can save time having a ton of meetings trying to find the right solution.  It’s easy to know a few vendors that might be out of your price range before even asking. Also, if you have a price in mind, they will show you the solutions to fit your needs.

Step 4 - Features.

Think about things from firmware updates, MFA on the storage, and other features you may need. 

 

Veeam works with anything from servers full of disks, to large NVMe all flash arrays. my workload involves many tape drives streaming at full tilt. having repositories and proxies with multiple 32Gb fiber and 25Gb networking in LACP was required to not saturate the links. There is always a bottleneck, but looking at your disk latency, networking and FC ports it helps to see what your requirements are as well. 

At the end of the day, disk is disk, but keep those things in mind while choosing. I use a combination and target the fast stuff for things that require performance, and the slower stuff for long term and size.