Concurrent Job vs. concurrent task

[klemeste]

Hi

I'm currently designing my new infrastructure based on Veaam 9.5. Reading best practices documents I decided to size my proxies regarding the potential concurrent tasks.
I'm wondering if a "task" is basically a job or a task is a VM processed in a job?
In other words, if my capacity is 6 simultaneous tasks can I run 6 jobs whatever the number of VMs or 6 VMs splitted in some jobs?
Thanks for helping

[DGrinev]

Hi Stephan,

Each task processes one virtual disk.
Please see detailed description in Veeam B&R users guide – Limiting the Number of Concurrent Tasks


Thanks!

[klemeste]

Thank you for your quick answer Dmitry.
Very interresting because my VMs got several vdisk each, I have to make the count.
The system requirement for a proxy server says:
CPU: modern x86 processor (minimum 2 cores or vCPUs). Using multi-core processors improves data processing performance, and allows for more tasks to be processed concurrently by proxy.
Memory: 2 GB RAM plus 200 MB for each concurrent task. Using faster memory (DDR3/DDR4) improves data processing performance.

The formula is clear in term of Memory but for CPU, I found this on expert db:
It is best practices to plan for 1 physical core or 1 vCPU each of the tasks.

Is it true for your point of view?

Thank you

[DGrinev]

Hi Stephan,

It is not mandatory but strongly recommended that you assign task limitation settings using the following rule: 1 task = 1 CPU core.
CPU resources are used for the data compression.

If I were you I’d setup backup proxy as recommended, then run a job and see if the bottleneck is proxy then you need to allocate more resources.

Thanks!

[klemeste]

Thank you Dmitry

Now from a Repository perspective:
The best pratice is to apply this rule:4Go memory + 4Go additional per "JOB".
So I have to ensure that all my chain is consistent: my concurrent tasks (processed by my proxy) must match the capabilities of my repository.
For instance:
My repositories are sized as: 16Go memory = capacity of 3 jobs (4 + 4*3)
My proxies can run 30 tasks simultaneously (defined by sizing).
I have to distribute the tasks following this formula:
- in 12 jobs if 4 repos (2.5 tasks per job)
- in 9 jobs if 3 repos (3.3 tasks per job)
- in 6 jobs if 2 repos (5 tasks per job)...etc

I not sure I'm clear

[DGrinev]

Stephan,

Memory is not the primary attribute for repository performance, since IOPs determinate how fast data will be processed.
Here is an article on memory usage by repositories.


Thanks!