Agent-based backup of Windows, Linux, Max, AIX and Solaris machines.
Post Reply
perjonsson1960
Veteran
Posts: 443
Liked: 44 times
Joined: Jun 06, 2018 5:41 am
Full Name: Per Jonsson
Location: Sweden
Contact:

Physical server agent backup throughput

Post by perjonsson1960 »

Folks,

I don't know if this only applies to our environment, but when I look at the throughput graph for an agent backup job, it looks like the teeth of a handsaw. It goes up and down in an even pattern throughout the entire backup. Is this due to something that the Veeam agent is doing, like preparing data locally in the backed up server, during which time there is no transfer of data over the network? I am also seeing this phenomenon when we are doing backup copies of agent backups from one physical Veeam server to another, residing at two different physical locations. In the latter case it can be a transfer of data for about 2-3 minutes, followed by 3-4 minutes with no transfer of data at all.

When we are backing up VMs in our VMware environment, we don't see this. There the data seems to be transferred smoothly throughtout the entire backup and backup copy.

Is this by design, or is there something fundamentally wrong with our environment?

PJ
DGrinev
Veteran
Posts: 1943
Liked: 247 times
Joined: Dec 01, 2016 3:49 pm
Full Name: Dmitry Grinev
Location: St.Petersburg
Contact:

Re: Physical server agent backup throughput

Post by DGrinev »

Hey Per,

Can you provide a bit more details, what data storage is used in the source server and repository, how many IOPS does both shows? Thanks!
perjonsson1960
Veteran
Posts: 443
Liked: 44 times
Joined: Jun 06, 2018 5:41 am
Full Name: Per Jonsson
Location: Sweden
Contact:

Re: Physical server agent backup throughput

Post by perjonsson1960 »

We have three identical HP DL380 Gen10 servers for our backup solution. Each server is equipped with an internal "disk box" with 12 disks and an external box with 12 disks. Each box is a repository, and the two backup servers' four repositories are added into one scale out repository, and the third server, used for archiving, has a second scale out rep. The backup servers are residing at one physical location, and the archive server is residing at another physical location, and there is a 10 Gigabit connection between the two locations. We are using RAID 5 and ReFS.

I am looking at the latest backup of one of the Exchange servers, and the average processing rate is 604 MB/s, and when I look at the graph, I get values like 5 GB/s read speed and 259 MB/s transfer speed. The 5 GB/s value is not physically possible if "GB" means Gigabyte. If it meant Gigabit then it would make more sense.

When looking at the graph, the periods with no transfer are longer than those where data is actually transferred.
DGrinev
Veteran
Posts: 1943
Liked: 247 times
Joined: Dec 01, 2016 3:49 pm
Full Name: Dmitry Grinev
Location: St.Petersburg
Contact:

Re: Physical server agent backup throughput

Post by DGrinev »

It can be related to the exceeded cache capacity of the Agent, since the reading speed is very high along with the pretty fast transfer.
So, in case where the cache is full it stops the transfer and doing only writing on the target side.

You can reach out the support team to find out the real cause of the issue by deeper logs diving. Thanks!
lperezg
Veeam Software
Posts: 82
Liked: 3 times
Joined: Nov 23, 2018 8:31 pm
Full Name: Luciano Perez Gaimberan
Location: MTY, Mexico
Contact:

Re: Physical server agent backup throughput

Post by lperezg »

Hi,

If the backup image is big or it's residing in a dedup appliance this could be happening due to the reading times - Veeam reads the data in the backup image to know what data needs to retrieve from the client.
Post Reply

Who is online

Users browsing this forum: No registered users and 7 guests