Calculation of bandwidth and Storage space

[zak2011]

A couple of queries:
1) Storage considerations for the DR site- When replicating VMs, will the replica be the same size as the original VM. Is there a simple way to calculate how much storage would be needed at the DR site if I need to replicate a VM (1TB) in size.
2) How do we calculate the amount of bandwidth required for replication if i do a daily replication? Is there a predefined way to do this?

Thanks..

[dellock6]

Can answer easily for the first one, since replica replicated a fully functional VM, its size is the same as the original one if you keep thick disks, or will be shrinked to thin disk if you choose so. But please also consider restore point saved as regular snapshots, they will use disk space too.
About n°2, since Veeam uses the same algorithms and optimizations for backups and replicas, the daily incremental size of a local backup is somewhat a starting point to find out how much data will be replicated daily. From here, you can further reduce this amount by using higher optimization (wan target and high compression).
Finally, fro here to define the needed bandwidth, we also need to know the duration you would accept to run the replicas

[Vitaliy S.]

There is an existing topic covering the second question, please take a look: WAN Bandwidth Requirement

[zak2011]

Thank you for the replies dellock and Vitaliy. To answer dellock's question on the duration we need is something between a 5-6 hour window.

[dellock6]

Ok, so as soon as you will figure out how much data you need to replicate, you will find out how much bandwidth do you need

Luca.

[zak2011]

Thanks dellock and Vitaliy for your guidance. As i am completely new to this, i may have a couple of queries coming up. Thanks for your time.

[zak2011]

As we are planning to have our DR site set up with Veeam, is there any sort of checklist that i should go through before implementaiton.
From my limited experience with Veeam, a few things that come to my mind
1) How many VMs will I need to replicate to the other site.
2) How many restore points would I need for the replication.
3) Based on how many replicas needed, how much storage would i require at the Recovery site.
4) Based on daily change rate, how much bandwidth would i require.

Is there any other factors i need to consider..
Thanks for your time!

[Vitaliy S.]

Your list looks good, though there are two things that I would add:

1. For low connection bandwidth consider using replica seeding option, which is available on the first step of replication job configuration.
2. If you're going to replicate offsite, do not forget to deploy two proxy servers (one at the main site, the second one at the remote site).

[zak2011]

Thanks Vitaliy for the suggestions. I will take this into account.

[zak2011]

Hi,
This may sound a little out of the way..but i would like to get this cleared.
1) Is it possible to have the deduped backup data ( Full Image backup + incrmentals ) both at the Production site and the Disaster recovery site by creating two backup jobs from the Veeam server at the production site. This way would i save space on the DR site since the original backups are deduped and compressed without doing replication and creating replicas
Hope this makes sense!

[foggy]

Yes, you can run two jobs against the same VMs to different backup repositories to have both local and remote backups. Alternatively, you can backup locally and synchronize the backup files to the remote site using some third-party software (to decrease the effect backup process puts on your production VMs), there are a number of existing topics on how this is done. However, take into account the time required to recover a VM from backup compared to the failover to its replica VM (which is, of course, much faster).

[zak2011]

Thanks foggy for the quick reply. Could you paste some links on the alternate suggestion you talked about. Will not Instant Recovery work from backup if powering on from the remote site.

[Vitaliy S.]

Here you go: http://forums.veeam.com/search.php?st=0 ... te+example

[foggy]

The most commonly used synchronization software among our customers is rsync. There are plenty existing topics on this forum about offsite replication of backups using rsync, you can use search and check them, as Vitaliy suggested. Here is the couple of the most recent ones: Best backup setup for mirroring backup respository and v6 - How to have a local and off-site backup copy?

Instant recovery is a perfect way to quickly start the VM from backup but it is not the right tool to use to recover the whole production site.

[zak2011]

Thanks for sharing the articles foggy and Vitaliy. Is it recommended to have the backup job one to local site and another to DR site along with replication? If i need to do this, what are the main things i need to take into consideration?
Thanks!

[Vitaliy S.]

Actually it's up to your DR strategy, but yes, it's better to have at least three copies of backup files (e.g. local backup files for FLR operations, remote backups for protection from source site failure and VM replication jobs for most critical VMs).

All setup tricks and tips have already been discussed in these threads, please check them out.

[zak2011]

Thanks for your recommendation. Could you send me the link you are referring to. I am not able to see it for some reason

[zak2011]

I have a couple of queries regarding Veeam. We will be going in for VBR6.1 Enterprise shortly, so would like to have clear understanding before we change from Backup Exec 2012 to VBR6.1
1) During the product demo of VBR6.1, it was mentioned about ‘quick migration’. We have a virtual infrastructure with vsphere4 standard. Do you mean by quick migration we could do both vmotion as well as svmotion even if we don’t have vsphere enterprise licensing?
2) Veeam ZIP- what is the main difference doing a normal backup job using Veeam and VeeamZIP. If I backup using VeeamZIP, does that mean I get more compression and dedup than what I get using normal Veeam backups.
3) Performance degradation may happen using vpower when running the vm from a backup. What kind of performance degradation? Could you be more specific about this? When it comes to an exchange server or an SQL servers or webserver..how much performance degradation should I anticpate?
4) Can I run the backup job both at production as well as the DR site at the same time? If so, will if affect the performance of the VM / proxy server at the production site? Do I need to have more proxies at the production for this? Also if I need to run replicas of perhaps just three or four VMs to the DR site , how should we design the number of proxies required at each site.
5) How do I anticipate how much bandwith I need if I am going to run a the entire backup job also at the DR site plus replicate a couple of VMS. What are the things I need to take into consideration or be aware of?
6) I have Backup Exec 2012 as my backup server. Is it allright to install VBR6.1 on this same server and install SQL as a separate instance. I will need to use Backup Exec 2012 since it is attached to a tape library and investing in another physical server just for Veeam at the production site is not a possible solution at the moment.
7) Also intelligent load balancing feature- Does this mean if I have proxy servers at each site, the load balancing will be efficient? How do I know which proxy is overloaded and which is not? Is there a way to see this?
When running a replication job, how different is it from a backup file? Is it just that a backup job has compression and dedup and the replica has only compression?
9) Are there any important things to consider before going in for release 6.1 ( any known issues to be aware of ).

Thanks!

[vMO]

I have a couple of queries regarding Veeam. We will be going in for VBR6.1 Enterprise shortly, so would like to have clear understanding before we change from Backup Exec 2012 to VBR6.1
1) During the product demo of VBR6.1, it was mentioned about ‘quick migration’. We have a virtual infrastructure with vsphere4 standard. Do you mean by quick migration we could do both vmotion as well as svmotion even if we don’t have vsphere enterprise licensing?

Veeam Quick Migration is a combination of our vPower and replication features and allows you to "Quick Migrate" a vm. if vMotion or SvMotion is avilable, it will be used, if not, replication comes in.

2) Veeam ZIP- what is the main difference doing a normal backup job using Veeam and VeeamZIP. If I backup using VeeamZIP, does that mean I get more compression and dedup than what I get using normal Veeam backups.

no, veeamZip is a adhoc fullbackup of a vm. without using cbt or other incremental features. there is no sheduling in the gui. so its wat it is, a adhoc vm fullbackup

3) Performance degradation may happen using vpower when running the vm from a backup. What kind of performance degradation? Could you be more specific about this? When it comes to an exchange server or an SQL servers or webserver..how much performance degradation should I anticpate?

depends on the storage your backup files are on. and on a bunch of other things like network, cpu, ram.. the best will be to test it in a poc if it fit's your performance needs.

4) Can I run the backup job both at production as well as the DR site at the same time? If so, will if affect the performance of the VM / proxy server at the production site? Do I need to have more proxies at the production for this? Also if I need to run replicas of perhaps just three or four VMs to the DR site , how should we design the number of proxies required at each site.

a vm can only be backedup once at the same time. a proxy needs 2 cpu cores (phys or virt) per task. if you configure your proxy to have 4 cpu cores, this proxy can run 2 parallel tasks. if you have two sites, it depends on the conectivity between, if it is needed to have a proxa on both sites, but perhabs it is. the proxy everytime has to be as near as possible to the source. we are doing snapshot based backup, the snapshot has to be open as short as possible, so the proxy who is dooing the work has to be as ner as possible, same with repository.

5) How do I anticipate how much bandwith I need if I am going to run a the entire backup job also at the DR site plus replicate a couple of VMS. What are the things I need to take into consideration or be aware of?

to figure this out, you can do a local backup in the same shedule, you will see the initial size and also the incremental size. we do incremental backup with cbt so it's block based, it doesnt care the amount of changed data, it cares the amount of changed blocks within this data. note, that we provide "replica seeding" to do the initial transfer by removable media, also we do a kind of forever incremental, so only changed blocks have to be transfered.

6) I have Backup Exec 2012 as my backup server. Is it allright to install VBR6.1 on this same server and install SQL as a separate instance. I will need to use Backup Exec 2012 since it is attached to a tape library and investing in another physical server just for Veeam at the production site is not a possible solution at the moment.

this will work, we only need the includes SQL Express from setup, since we do not store that much data into the SQL

7) Also intelligent load balancing feature- Does this mean if I have proxy servers at each site, the load balancing will be efficient? How do I know which proxy is overloaded and which is not? Is there a way to see this?

you will se that in the backup statistics. inteligent load balancing means the loadbalancing of jobs between proxys, to do that work, datasore connection and load and others will be counted.

When running a replication job, how different is it from a backup file? Is it just that a backup job has compression and dedup and the replica has only compression?

the only diference is in the target side, a backup job is stored as a vbk/vib file on a repository, a replicaktion job is stored as native vmdk/vhd file on the hypervisor. compression and deduplication will still be used during the transfer, since the transfer goes from proxy to proxy by a replication job. backup will store also the backup files in a deduped/compressed format.

9) Are there any important things to consider before going in for release 6.1 ( any known issues to be aware of ).

make sure read the release notes and if you are dooing an upgrade from 6.0 to 6.1 first update enterprise manager.

Thanks

[Vitaliy S.]

Could you send me the link you are referring to. I am not able to see it for some reason

Here is one of the existing threads that specifically talks about WAN replication techniques: Replication from HO to DC

[zak2011]

Thanks VMO for the answers. Great job! For no.4 I understand from your response the proxy has to be near the source inorder to do the job. We will be purchasing an EMC VNX at the primary site and shift the exisiting HP SAN to the secondary site. However i will be having another repository ( exisiting SAN) at my DR site to which the proxy at the production site will be directing the second backup job. Is this okay? I will have fiber link between both the sites. For no.5- If i am going to do a Full backup every week and then incremental backups everyday, i am concerned about the Full backups. If i use forever incremental then i wont have to worry about this. Currently my backup plans take full backup every week and then incrementals every day. Are there any best practices with respect to the Full backup from Veeam's point of view?

[foggy]

However i will be having another repository ( exisiting SAN) at my DR site to which the proxy at the production site will be directing the second backup job. Is this okay? I will have fiber link between both the sites.

If you are talking about having two jobs writing to different repositories, then this is ok.

For no.5- If i am going to do a Full backup every week and then incremental backups everyday, i am concerned about the Full backups. If i use forever incremental then i wont have to worry about this. Currently my backup plans take full backup every week and then incrementals every day. Are there any best practices with respect to the Full backup from Veeam's point of view?

If you're planning to perform weekly synthetic fulls (backup will be built synthetically from the previous full and subsequent incrementals), then this activity takes place locally on the backup target repository and does not affect the network. Active fulls (created by pulling the VM data from the production storage) could be scheduled monthly or even more rare.

[zak2011]

Thanks foggy and Vitaliy for your help and expertise. More queries to come when i work along with the product.
Thankyou very much!

[zak2011]

I must also say the level of explanation is really good on the veeam forums. Its really cool to see the Veeam experts can explain things in simple terms to novices like me
Thanks!

[davidb1234]

[merged]

The replicated VMs will be stored on a FC SAN at the destination DR hotsite connected by 150Mbps WAN link.

How much disk space will the Veeam Proxy server need at this location? We have approximately 30 VMs for a total of 3.5TB of data.

I am thinking Veeam doesn't need much space though. I've been looking through the documentation and it doesn't appear to need much but want to make sure I am not missing anything.

The Veeam Proxy server will be a repurposed server with 8 cores and 50GB of RAM running windows 2008 R2. It only has mirrored 72GB disks though and I want to make sure that will be enough.

[dellock6]

Hi David,
since replicated VMs are written in the same format as source, the required space will be the same you are using now, so 3.5 Tb for the first copy. After this, you will also need additional space for the retention points, saved as snapshots in the destination datastore.
One important thing: if you are using thick disks at source, you can save much space by converting vmdk to thin at DR site, this is a great feature of Veeam.

Luca.

[davidb1234]

This is just on the SAN though right? The DR site Veeam Proxy server itself doesn't need much space just for the software installation and SQL database right? Or does it need a large amount of space to stage data before copying to the SAN?

[foggy]

Yes, the amount of free space on the backup proxy itself does not matter. It does not even have the database installed on it (unless it is the default proxy, which is a part of the Veeam B&R installation). What is more important, is the amount of CPU cores on the proxy server as all the processing load is handled by the proxy.