20 | Veeam Backup & Replication | VMware Environments | USER GUIDE | REV 1
writing data to target. VM data therefore is passed over the data pipe. When one data processing
cycle is over, the next cycle begins.
Veeam Backup & Replication provides advanced statistics about the data path performance and
lets you identify bottlenecks in the data transmission process.
To evaluate the data flow efficiency, Veeam Backup & Replication analyzes performance of all
components in the data pipe working as the cohesive system, and evaluates key factors on the
source and target sides.
(1) Source – the source disk reader component responsible for retrieving data from the
source storage.
(2) Proxy – the backup proxy component responsible for processing VM data.
(3) Network – the network queue writer component responsible for getting processed VM
data from the backup proxy and sending it over the network to the backup repository or
another backup proxy.
(4) Target – the target disk writer component (backup storage or replica datastore).
The resource usage level for the four points is evaluated in percent. This percent rate defines the
amount of time for which components are busy during the job. When you run a job, the VM data is
sent continuously over the data pipe to the destination, block by block. An efficient data flow
assumes that there is no latency at any point of the data pipe and all its components work for
approximately equal amount of time. If any of the components operates inefficiently, there may
appear a bottleneck in the data path. The insufficient component will work 100% of time while the
others will be idling, waiting for data to be transferred. As a result, the whole data flow will slow
down to the level of the slowest point in the data path, and the overall time of data processing will
increase.
To identify a bottleneck in the data path, Veeam Backup & Replication detects the component with
the maximum workload – that is, the component that works for the most time of the job. For
example, you use a low-speed storage device as the backup repository. Even if VM data is retrieved
from SAN storage on the source side and transmitted over a high-speed link, VM data flow will still
be impaired at the backup repository. The backup repository will be trying to consume transferred
data at the rate that exceeds its capacity, and the other components will stay idle. As a result, the
backup repository will be working 100% of job time, while other components may be employed,
for example, for 60% only. In terms of Veeam Backup & Replication, such data path will be
considered insufficient.
The bottleneck statistics for a job is displayed in the job session data. The bottleneck statistics does
not necessarily mean that you have a problem in your backup infrastructure – it simply informs you
about the weakest component in the data path. However, if you feel the job performance is low,
you may try taking some measures to resolve the bottleneck. For instance, in the case described
above, you can limit the number of concurrent tasks for the backup repository.
Backup
Unlike traditional backup tools designed to work with physical machines,
Veeam Backup & Replication is built specifically for virtual environments. It operates at the
virtualization layer and uses an image-based approach for virtual machines backup. To retrieve
VM data, no agent software needs to be installed inside the guest OS – instead,
Veeam Backup & Replication leverages ESX snapshot capabilities. When a new backup session