Databases Reference
In-Depth Information
Fig. 3.
System for Deploying Virtual Desktop
Thus, in accordance with the proposed method, the process for deploying a
large number of virtual client environments is automated.
4.2 Proposal of Load Balancing during Use
The measures mentioned above are a proposal to simplify the initial settings for
a large number of users. In this section, we propose load balancing for when we
apply various processes such as updating the OS, the security patches, and virus
scans in multiple numerical virtual desktops after the deployment.
Load Balancing by One to One Grouping.
Figure 5 is an example of load
balancing by mapping a blade and a storage unit one to one.
We assign multiple (e.g., 40-60) virtual desktops on one hypervisor to an
individual datastore. Thus, virtual desktops on different hypervisors do not share
the same datastore. Because a virtual desktop on a certain hypervisor is not
affected by the virtual desktop on other hypervisors when we update the OS
and application or do a virus scan, we can realize load balancing.
However, for example, this assumes there is an obstacle to the blade server on
which there is one hypervisor and that the blade server is restored by the HA
configuration (dualization) automatically. In this case, the problem remains that
the load concentrates on one storage because disk I/O occurs from all virtual
desktops on one hypervisor at the same time.
Load Balancing by Meshed Grouping.
Figure 6 is an example of load
balancing that improved the configuration that we showed in figure 5.
We disperse and assign each virtual desktop on one hypervisor to multiple
datastores in order to plan for dispersion of the network load between a hyper-
visor and the datastore in this proposal. Unlike the configuration of figure 5, the
Search WWH ::
Custom Search