Managing complex VDI environments: Simplified with Parallels RAS

Parallels
74 followers
Mar 11, 2026
This session is built for advanced technical users, including architects, senior system administrators, and security officers who work with complex VDI and application delivery environments.
Hosted by our Sr. Sales Engineering Director, it explores how to manage multi-data center and multi-tenant setups while keeping costs down and performance high.
The session features content from the Advanced Architectures module of the Parallels RAS technical professional advanced training, giving you practical guidance you can apply right away.
AI Generated·May contain errors
Good morning, good afternoon, good evening, depending from where you are joining us today.
My name is Steven de Winter, I'm the Senior Director for Sales Engineering at Parallels
and I will be your technical host for today's session. I have a broad background in all kinds
of IT infrastructure, cloud and user computing, so this is not going to be your typical
marketing webinar from Parallels, but this is going to be a technical session in which I will
go through some topics which we also typically cover during one of what we call our technical
professional advanced training. So what we're going to discuss today is how ROS architecture
looks like, but not only the simple ones, but also I'm going to go into details how you can
from those simple designs and those simple components also build out a ROS environment
or a Parallels ROS environment which is like enterprise ready. So that means that we're going
to discuss what makes an architecture enterprise ready. We're going to discuss concepts like
multi-site deployments, multi-tenant scenarios. We're also going to have a look in how you can
set the Parallels ROS covering multiple data centers and then depending on certain scenarios,
what is the most appropriate architecture decision. And then linked to these enterprise
scenarios, we're also going to discuss some of the more advanced features like preferred routing,
multi-hop architectures and then also from a client point of view like okay your backend and
your infrastructure is then enterprise ready, but how can you deploy your clients also in an
enterprise context. And we will have plenty of time at the end of the presentation today for a
Q&A. So as you can see in this GoToWebinar, there is a possibility for questions and answers.
So this is a live webinar, so if you have like any questions, don't hesitate to put them
preferably in the Q&A part of the GoToWebinar or in the chat. At the end of the presentation,
I will have a look at the Q&A part and try to answer as many questions as possible.
But let's start. So let's maybe start with a one minute quick recap of the Parallels
architecture. So for those of you who are not super familiar with Parallels ROS,
Parallels ROS has multiple components and the main components of it, the key components,
is what we call the secure gateways and the connection broker. So these are the mandatory
components that you need in any infrastructure, in any design or in any implementation you do
from a Parallels ROS. The secure gateway is the component which is used for the clients to connect
to and the connection broker is let's say the brain of the systems and it's a connection broker
that then for example based on the user identification then for example will list
which applications or desktops or maybe even physical PCs that a certain user can connect to.
Besides the secure gateways, you can scale them out. You will see that later in today's session.
You can have multiple secure gateways for scalability purpose. If you need to make your
secure gateways also high available, we have a component called the high available load balancer.
That's also a component that comes from Parallels that you can put in front of your
gateways to make them high available. So typically either you're using your own
third-party load balancer. So for example if you're deploying on Azure or any of those
cloud platforms you can use an existing cloud load balancer from the hyperscaler or if you're
deploying on-premise you can for example use the built-in high available load balancer from
Parallels. From there your users will then be load balanced to one of your secure gateways.
On the secure gateways you're connected them to the connection broker. The connection broker
will do the identification, will list the applications and desktops which are available
and then the sessions can start. So this is the main concept and the main architecture components.
Something important that Parallels also does is it has integration points into underlying
hypervisors or cloud platforms. We call these providers and at the current time of this
webinar we have seven providers. These are for on-premise Hyper-V, VMware, Nutanix and
scale computing. So if you're running in your own data center you can integrate into these platforms
and do things like auto scaling and power management. If you're using cloud deployments
we have the same functionality for Azure, Amazon EC2 and then also for Azure Virtual Nest. So
if you're using any of these platforms Parallels secure, Parallels remote application server
can also automate and scale out your environment. With these components
recapped and rehearsed let's have a look at what we would call like a typical simple
architecture and if you take a look at the at the single farm single site architecture
that is a deployment or an architecture type where we can already make everything high available.
So this is not something complicated for those of you who already have done Parallels installations.
Whenever you start your installation you typically start with deploying your private
connection broker. Once your primary connection broker is up and running and it's basically like
next next next finish install you activate either the built-in license or you activate it
with your own license that you have purchased. Once you're in your primary connection broker
via the Parallels console or via the management portal from there you can do a full deployment
of the other components and as you can see this is what we call a single site single farm setup
we're going to discuss later on more advanced architectures but let's say in a single site
single farm setup it is perfectly possible to have all the components high available. So for example
what you can very easily do is for example deploy a secondary connection broker
and then whenever your primary connection broker for example needs to be rebooted or is down or
is broken the secondary connection broker can take over or for example for scalability if you're
going above a certain amount of connections you could have like multiple connection brokers to
load balance the traffic and there's the same story with the secure gateway so we can have
multiple gateways where then for example the high available load balancer is in front of it for
making things high available and the high available load balancer that's a virtual
appliance so for those of you who are not super familiar all other components or
software which is deployed on top of Windows servers but the the HAL-B or the high available
load balance that's a virtual appliance that you can spin up on your own infrastructure
of course for if you want to make things high available and you only have like one
HAL-B then you still have like a single point of failure in your architecture so you can have
multiple ones and then those can be made high available by adding a virtual IP so you can assign
a VIP that then load balances between those HAL-B so technically speaking with your clients
you connect to a VIP IP address which balances between your HAL-B load balancers then that
HAL-B will pick any of the secure gateways and then connect you for example with your desktop
or your session host in the in the back end. This scenario scales pretty well we can
easily go up to 5,000 concurrent users that's not an issue of course
if you go to 5,000 concurrent users you might need for example maybe like four or five or six
secure gateways and you might need like not two but three connection brokers but
it's a very simple and linear model where you can by adding more gateways and more brokers
up to 5,000 concurrent users. Also the 5,000 is not a hard limit it's a sort of recommendation
you will see when we're discussing more advanced architectures later in this webinar
we're going to show for example how you can go to like a 10,000, 20,000, 50,000 concurrent user
scenario. Another thing which is important in a single site is that the connectivity between
the connection broker and the other components is fairly fast and what do we mean it's fairly
fast that we see like a latency less than 50 milliseconds. So if you have two data centers
and they are well connected then typically that interconnection is less than 50 milliseconds
that's fine you can set up like a single site across two data centers if you have
and that latency is good but once the latency goes above that 50 milliseconds
you will see that we have other architectures that we would recommend. So 5,000 it's not a
hard limit but if you go above it we should consider looking at making some modifications
to the architecture and the same thing with the latency of the network. Even in a single site
it is technically possible to do for example user segregation so there is a concept which is called
teams you could for example create like multiple teams and then you could say like customer one
connects to team one, customer two connects to team two, customer three connects to team three
and then you could give for example every customer like its own look and feel it's
their own policies, their own service, their own application so that there is an easy way to
also segregate users into even a single farm. Very important that is that in a single site and
also in a single farm it is also important to have all the components at the same version and so that
means that for example today most customers will be like on version 22 we made version 21
available since yesterday so if you go to the parallels my account you will see that the
parallels 21 is actually available it's GA it's general available since yesterday. We didn't send
out the marketing emails yet so for those of you who are joining me today during this live webinar
you get like a sort of scoop version 21 is GA it's available for download in my parallels.
If for example you want to upgrade your existing environment to the new version it is important
that you upgrade all the components more or less at the same time so that means that
once your connection broker is updated to version 21 that you also upgrade pretty fast
your secondary connection broker that you upgrade your gateways and so on and so on.
Important also to mention is that during the upgrade process for example of a single site
if for example you need to upgrade your connection brokers that can be done without
user impact and if you need to upgrade your secure gateways and these require typically also like a
reboot it is possible to do that so you can for example upgrade your first secure gateway
at that time users which are connected to your first secure gateway will be like prompted that
the connection is dropped and then after a few seconds they will be automatically reconnected
to the to another secure gateway and then you wait until the first secure gateway is back and
you do the same for the second secure gateway so there is a fairly simple process to upgrade
everything to the new version in a single site. The only thing which is important is like that
you don't put like days or weeks in between that once you start with the upgrade that you
continue with the rest of the infrastructure at the same upgrade moment. Now if we take a look
at these limitations and for example what if you want to go above a certain amount of users or what
if your network is not good or what if you want to do like full isolation or full
disaster recovery then I think it's important that we take a look at some key considerations
for let's say going like one step further into like enterprise designs and it's not only about
for example scalability and scalability is something pretty obvious if you go to 10 000
or if you go to 20 000 users you have to make another architectural design then if you deploy
this for 20 or 50 users but it also has to do with stability and with expectation
management. For example what if your company policy is that you
have for example two data centers and that you want to use like one data center active and the
other one passive or that your policy says that you need to upgrade first your one data center and
you can only touch your secondary data center for example two weeks later in another
maintenance window so these are all considerations that have to be taken into account
when designing what we call these enterprise architectures and one thing is key and you
will see that during today's presentation that simplicity for me is very essential so
where possible we will not try to make things more complicated than needed. Before we go into
these multi-site and multi-talent deployments and architectures I think it's important to
explain a few concepts and also take a look under the hood on how certain things work in parallels
and this information I'm going to share is something you typically will not find in any
public documentation it's some of the internals and I think it's worthwhile for
these kinds of more advanced webinars to also give you some insights on how the product works under
the hood. So there are basically three layers we have the access layer, we have the logic layer
and the connectivity layer. The access layer is what we call like the secure gateways and the
high available load balancers then the connection broker and the enrollment service what we call
the logic layer and then the connectivity layer is the agents that run on for example the session
host and the VDIs and the drawing next to it is a little bit misleading because there is not
such a thing as for example like a configuration a shared configuration database some of our
competitors have like a SQL database that is used to store the configuration settings that is not
how it works on parallels. In RAS what you will see is that the configuration itself is stored
on the connection brokers itself so each connection broker that you have has like a copy
of the configuration and it's actually there are some mechanisms in place that for example
replicate these configurations to the other connection brokers and this is also
a little bit explaining why certain ports need to be opened, why certain considerations are
to be done and this will also explain later on when we're discussing multiple scenarios on
how for example we can do upgrades or why a certain design is better than the other designs.
So the first thing you need to know is that the primary connection broker of the first site
has a special role it is actually the one that has like a sort of like read-write access to the
configuration and whenever you make like a modification to a configuration
I think you remember that if you're like in the console you have to click that apply button
actually it's on the moment you click that apply button that the primary connection
broker will via the redundancy service push the configurations to the other connection brokers
and then for example once it's over there those other connection brokers will then
for example talk to the provider agent and then the provider agent will for example start
communicating with the VDIs or your session hosts or your systems that run your workloads
and so that's a bit how it works so you make a modification on a primary connection broker and
then the connection broker will first replicate it to the controller service on the other
connection brokers and then from there it will talk for example to the provider agent if there
needs to be like any communication with the other agents on the connectivity layer.
If you take a look on the connectivity layer we have multiple agents for example and it also makes
a difference if you're using like templates or static hosts so you will for example see if you're
for example using static session hosts that you're only using the RDSH agent but for example if
you're using templated session hosts that you're also going to use the guest agent so the guest
agent takes for example care of everything which has to do like with autoscaling and provisioning
and templates and clones and then for example the RDSH agent that is for example then taking
care for the specifics linked to the multi-user windows operating
systems like server 22, server 2025 and so on. In a multi-site setup and I'm going to explain what a
multi-site setup is in the next slide. In a multi-site setup it's also the primary connection
broker of the primary site so the first site that will first replicate the changes to the
connection broker of the secondary sites and then the primary connection broker of the secondary
sites will do then the replication to the provider agents and the system so that's
also I think something important that you need to know how these things work.
I'm not going into too much details on this one but I think it's important to
highlight that if for example you're deploying Parallels RAS that because of that replication
system and because the fact that those connection brokers need to communicate independently from
each other that there are like different network flows which are in use so we have
of course on one hand the pure network flows which are linked to the user traffic and then
typically for example from the client through the DMZ to the secure gateways for example
which is like typically 443 HTTPS traffic and then from the secure gateways we go via the
RDP protocol 389 to the session hosts or to the VDIs and so that's the
let's say the traffic in gray on the top but all these components also need to talk
to each other and these use other ports so it's important that if you're making for
example an architecture and a design and you're working for example with multiple sites and
you have for example a site in your first data center and a site in your secondary data center
then it's for example also important that next to that user traffic that also some of those
communication flows and communication ports are open and especially for example if you start
integrating into these providers and these providers have like different ports for example
like if you integrate into VMware or scale computing just to pick two of them it's based
on 443 traffic but for example if you integrate into Hyper-V it's another set of ports that need
to be open so if for example you have a host pool in your first data center you have another host
pool in your secondary data center then make sure that all your connection brokers can for example
talk to all these ports and all these providers when it comes to like designing your
network schemes. Besides the single farm and single site which we have
discussed already before one of the possibilities with Parallels RAS is
in your same configuration so in your same farm is to deploy a secondary site or a third site
or a fourth site so you can in a single farm deploy multiple sites and why would you do that?
That could be for example because you have for example multiple customers that
you're hosting and you want to isolate those customers on the site level or this could be
that you have for example a site data center A and a site data center B and if you want to isolate
certain resources like gateways brokers from each other and make them completely isolated
introducing like an extra site is one of the possibilities that you could do. This could also
for example help you scale about 5,000 concurrent users you could for example create a site
for your first set of users and then create a secondary site for your second set of users
and then you could for example partition your users you could for example say like these
these 3,000 users connect to that site these other 3,000 users for example connect to the
secondary site. Important is that all the sites run isolated from each other so
the gateways in a site only can access resources and connection brokers in the same
in the same site. What is important is if you're using like a single farm multi-site setup
and you have for example two sites you're still consuming one license. What do I mean with that?
It's still your primary connection broker in your first site that will be your licensor
and that also means that for example if the same user let's for example assume that that your two
sites are connected to the same active directory and that as an admin you log into your first
site and at the same time you log in into your second site and that user will only be counted
towards one license on the parallel side so that's I think something important to mention
and what is also possible between sites is the replication of settings and applications so
you could for example when you publish like an application in your primary site
you could make automatically that it's also published in your secondary site. You see the
asterisks next to the replication of sites so we replicate a lot of settings but for example
something we don't replicate is the host pools so if for example you publish an application it
will be also published in your secondary site but then you still need to link it to the host pool
in your secondary site because like host pools are typically concepts which are like isolated
in a site so you cannot link from one site to a host pool in another site that's for example
something that is not possible. Another thing which is also still linked with that concept of
single farm multi-site setup is that all your sites must run the same version of ROS
and that has to do with the previous slide where we mentioned that for example if you do
like a configuration change it will always be done on your primary connection broker of your primary
site and then that configuration change is replicated to your other sites so having a single
farm sorry having a single farm with multiple sites will help you for example with the latency
it will help you with the 5000 concurrent user limitation but it will not help you for example
if you still if you need to have for example one data center on version 20.2 and upgrade your
other data center to version 21 so if you want to go into these kinds of scenarios then
we have to look at multiple farms and we need to set up multiple farms completely isolated from
each other and then we could for example use our API to replicate or some splits to replicate
certain settings or certain configuration between the different farms. Important when you switch to
multiple farms you will also have like multiple primary if you switch to multiple farms you will
have multiple primary sites and if you have multiple primary sites you will have a multiple
primary connection broker in your first site so that also means that you have multiple
license servers so if for some reason you end up having multiple sites and then sorry multiple farms
then you will also need multiple licenses so you will for example need to chop your license into
sub licenses or if you're a splugger you will have to create multiple sub licenses in your
subscription. A special form of multiple farms is what we call multiple tenants
so Parallels RAS has a has a concept which is called multi-tenant and multi-tenant is a scenario
where you have multiple farms but where you can share the same access layer so you could have a
shared access layer in what we call a tenant broker so all your connections go to a
shared access layer and then from that shared access layer you could for example go to a tenant
one which would be like customer one in the customer's data center and then maybe go to a
tenant two in another data center or maybe on azure or maybe in your own partner data center so
we can put multiple farms behind what we call the tenant broker and then have a multi-tenant
setup. What are the advantages of such a multi-tenant setup? One of them is cost savings
you could for example make a full high available access layer and then have maybe like smaller
customers hosted behind maybe with a single server access connection broker and session
host maybe have like hundreds of like five to ten user customers you don't want to have like
separated gateways which are internet facing separate connection brokers you could for example
make a scenario where you have like for each of your smaller customers like a dedicated
connection broker session host but then have like a multi-tenant setup multi-tenant broker
for the access layer or if you want to have like only one IP address on the outside world
but have like multiple isolated environments on the inside world that could be another
reason why you want to set up these kinds of environments. So multi-tenant is something that
comes also with the Parallels RAS. You don't need a license on the tenant broker
environment so if you deploy the tenant broker environment you don't need a license on that one
but you would then need multiple licenses on your different tenant farms that you put behind this
tenant broker. Very practical so if you're not familiar with setting this
tenant broker up if you deploy Parallels RAS and you run the installer you will probably recognize
this screenshot on the left side so if you want to deploy like a standard setup or you want to
install like a tenant that you can then later on integrate into the tenant broker you go for the
first option in Parallels RAS remote application self-deployment or the custom one. Very specifically
if you want to deploy the tenant broker you have to go for that the middle one the Parallels RAS
tenant broker deployment and then it's like a next, next, next finish and you can install. The
tenant broker could for example be helpful if you want to go above 5,000 concurrent users you could
still have like one single entry point into your network but then you could have like multiple
tenants behind it that each go to a certain part of the road. We already spoke about the licenses
so I'm not going to repeat it so there's no license on the tenant broker but if for example
the same user logs in at multiple sites at the same time they will be counted for multiple
license. It's also important that and I'm going to show that on one of the next slides
the decision of which tenant is opened or to which tenant the user is
sent is based on the host header of the connection so if you want to have like
multiple tenants behind the tenant broker you only need one IP address
but you need different DNS names so you need for example customer1.mycompany.com
customer2.mycompany.com and then based if the customer connects for example customer1
the gateway the tenant broker will know that needs to be redirected to that tenant or if
connects to customer2 that it needs to be redirected to the secondary
tenant broker in this scenario. Also important all the authentication happens on the
connection broker in the tenants itself so nothing happens on the tenant broker so technically
speaking your tenant broker could be in a workgroup it doesn't have to be
a domain joint or if you're like doing hosting for multiple customers you don't need to
have like complex scenarios where you have like
forests and domain trusts and all these kinds of things everything can happen
completely isolated from each other. So in a multi-tenant setup we're sharing the
access layer but the other layers the resource layer and the control layer are completely isolated
from each other. Another very practical point so how do you then okay you have your tenant
broker setup so how do you then integrate those other tenants. You can integrate an existing
environment so if you already have like a deploy the standard Parallels ROS environment
there are two ways that you can integrate that existing environment into the tenant broker.
One way is with an invitation hash so as an administrator on the tenant broker you can create
an extra tenant you can specify everything it will create a unique invitation hash you
send the invitation hash to your customer or you log in yourself into that customer
environment you go to join tenant broker you type in the invitation hash and that
environment will automatically be integrated into the tenant broker. The other way it can work is
with a shared secret key so instead of like pre-configuring all the tenants that can join
you can also create like a sort of unique secret key and then you go to your existing environment
and then you say like join and you use that secret key and then it will automatically register the
the farm into the into the tenant broker so you don't need to prepare anything up front
it will automatically integrate it and put these farms together. Once a farm has been
joined into a multi-tenant setup you can also still disconnect it so you can also disconnect
from the tenant broker and another thing which is important is like if you're joining a
farm into a tenant broker it's not needed anymore to have secure gateways in the
in the tenant itself it's because you're using the secure gateways from the shared tenant broker
but it's also not making it not possible so you can still have for example also a secure gateway
in a tenant which is joined into a multi-tenant setup. Why would you for example want to do that?
Maybe you're I'm saying something like a service provider for a healthcare provider
and you have a tenant broker setup your tenants are in the hospital itself it might be
interesting to have for example a local gateway so that local connections can connect to the
to the local gateway but users are coming from the internet have to pass your tenant broker
gateways where you're adding for example third-party security services. So very simple
to set it up you deploy the tenant broker and then you might already have
existing environments that you can then join via those invitation hashes or via those
via the shared secret. Invitation hash can be used once needs to be pre-configured
the secret key can be used as many times as before and doesn't need to be
pre-configured. And if you then want to take a look how it how it works from a from a logical
point of view so for example user one connects to tenant one and the load balancers picks one
of the secure gateways the secure gateway talks with the connection broker in the tenant broker
it recognizes that tenant one needs to be redirected to farm one so what will happen the
gateway will loop all the traffic through this this farm the connection broker here will be used
to do the user authentication and then once the user is authenticated he can then for example
start start applications. Things like teams are all managed on the on the tenant farm itself
and or made then automatically available through the through the gateways in the
tenant broker so this is not something you have to worry about these things are automatically
replicated. Certificates that's another thing if you're using certificates these certificates are
managed onto the the tenant broker setup so either you upload the certificates over there
or you're using let's encrypt from the from the tenant broker to fetch for the certificate but
the certificates is one of the only things which is managed from the from the tenant broker itself.
So if we if we combine everything together then we end up with this kind of
of matrix where we have the four different architectures in single farm single site
single farm multiple sites multiple farms or the tenant broker with multiple tenants
and then we have like certain criteria if you go about 5 000 concurrent users
again it's not a hard stop it will not break on a single farm we generally recommend that
that you look at other scenarios for example in in that case or for example if you if you
want to do like full disaster recovery and it's you can do that with a single site single farm
but it's for example more recommended to use other other setups for for doing doing that and
let's let's take a few examples instead of going over the matrix let's let's just take a few
few examples first example could for example be you have multiple data center and your customer
wants to have like an active passive setup that can be activated in the case of a disaster recovery
one of the possible recommended scenarios and of course you always have to take a look in
every every case but one of the possible scenarios could be that for this kind of
requirements you you do a single farm with multiple sites so that means that you that you
create a site data center one a site data center two you make sure that there is network in in
between you make for example your primary data center your primary site your secondary data
center is a disaster recovery site one of the advantages of this kind of scenario is that
for example whenever you publish an application in your primary site it will automatically also
be like published in your secondary site so that that will that will make make sure that you're
seeing that your that your configurations between your sites are more or less synchronized of course
you still have to link it to your your to your pools but that's that's something that can be
maybe scripted or or be done fairly fairly automatic and then what you could do is for
example in the event of an of an incident you could configure your client to either connect
to a load balancer ip that you that you switch or you could for example also configure your client
to have like alternative or secondary connections so you can for example configure your client
saying like this is my parallel environment this is the primary connection parameters but you could
for example also specify like a secondary connection and if for example the primary
connection is not available anymore your client will be automatically redirected to your your
secondary connection and that secondary connection could then point to your secondary site and allow
the users to transparently connect to the secondary site for example in the event of a
of a disaster. Important and I'm going to repeat it again that that both sites need to be at the
same RAS version so if you upgrade your primary site you need to upgrade your secondary site
pretty soon after that because the because there is a replication between the sites
those settings and those parameters need to be the same. And let's have a look at the second
second example a second example is for example two data centers active active and there is a
very good very good network in in between and let's assume for for the ease of the exercise
that that it's less than 5 000 concurrent users and this could be for example a nice scenario where
where you don't make things more complicated than needed you just deploy like a single site
across across the two data centers you're putting some secure gateways in your first data center
you're putting some secure gateways in your second data center you put some gate some connection
brokers in your first data center you put some connection brokers in your second data center
and and you let your your network layer manage the public IP inflow. I'm assuming if you have
these kinds of of data centers that you probably also have like mechanisms where you can route your
traffic to one data center or devote your traffic to another data center so so you can play for
example with public IP addresses that are not public IP addresses but IP addresses that can
jump from one data center to another in the unlikely event that there will be like a full
full disaster. One of the main advantages is that that that you don't need to do any changes like
you're using your data centers active active so that that means that all configurations everything
is in use. You could for example use the auto scaling feature if you're sending for example
more traffic to to the host pool in one data center that automatically scales up and takes
and for example it's that if you have for example five servers in one data center five servers in
the other data center another host pool your primary data center goes down all your traffic
is rerouted to your to your secondary data center you're hitting for example the 80 percent load on
your on your host pool in your secondary data center then you could for example use the auto
scaling and parallel trust to auto scale that that pool up to a higher level to have more more
resources and one of the advantages in this scenario is that you don't need to do any
replication that everything is done automatically. In this scenario and also in the previous one
take into account for example that if your primary data center would be out for longer than 72 hours
that you will have to take some actions to for example promote another site or another
connection broker to primary connection broker for licensing
validity reasons but besides that it gives you like time to to react and it's only like
within 72 hours that certain actions need to be taken so it gives you like plenty of time to
to react. A third example is a multiple data centers but with not a shared network
for these kinds of scenarios we would recommend to to either go to like a multi-talent setup or to
a setup where you're having like two completely isolated farms on the positive side that would
allow you to do for example upgrade of one site before upgrading the the other side and this is
also something I didn't didn't mention but for example in a multi-talent setup you can have like
sites on or farms on different levels so you can have like one farm which is connected to
your multi-talent environment on version 21 and you can have another farm at the same time
on version 20 that's not a problem. On the negative side you would need to take care of
replication as mentioned before so you probably might need to do some scripting or some
processes in place to make sure that your two sites are the same and also also from a
license point of view you will have two licenses so typically for example if you're like an m
customer and you have a 500 user license and you're doing like an active active over two farms
you'll probably have like split up your license into like 250 250. On the short term there will
be no issue because parallels allows you to do like over over subscription for a short time so
like if all of a sudden your 500 users end up on your on your 250 user
licensed environment that's not going to be a problem that will work but if that
situation remains there for a longer time you will have to take some actions on your
on your license and for example decalibrate your license to have more licenses in your
primary data center and then in your active data center and then for example less licenses in your
data center which is which is offline but also on the positive side this is like a real true
let's say freedom besides architectures there is also a few other concepts that I would like to
cover in this advanced webinar one of them is preferred routing so what is preferred routing
preferred routing is a feature in in parallels that allows you to pick a gateway based on the
target resource so let me explain it with an example let's say for example you have a hybrid
deployment where you have certain resources running in your on-premise data center and you
have some resources running in in your cloud data center let's for example say azure and you have a
gateway in your data center you have a gateway on on azure and your customers connect randomly
so technically speaking it could be for example that one of your customers connects to your gateway
in your in your data center and he needs to start for example a desktop which runs on on azure
and what we could do on that desktop on azure we could add a preferred route on that desktop on
azure and say that if this resource is started the preferred route for this desktop is to use
the gateway on azure and not the gateway in the data center because if you would do that then all
the traffic would need to go from azure first to your to your gateway in your data center and then
go to the outside world so that would for example put quite a that could for example put quite a
lot of stress on your on your network between for example your data center and your and your cloud
so so preferred routing has nothing to do with which gateway the authentication is done with
everything to do with which gateway is used for starting starting applications so very practical
if you want to if you want to set this this up you can go to any of your published applications
or to a group of published applications you can click on the on the routing tab and then based
on the on the routing tab you you can add preferred routes these roads could be for example be like
how they load balancers this could be secure gateways or this could be like custom
entry points on your on your network and maybe maybe you have maybe you're running on cloud and
you have like an availability set in in asia you have another availability set or another sorry
another gateway in in europe it might make sense that that for example an asian desktop is sent
out via the the gateway in the in in asia and that you for example using like only the load
balancer to to do the correct to do the correct roping so you can play with all these objects
and then you can make a list and can say for example the preferred road is first is then it's
that and then it's that so if you take a look at the at the algorithm and in this example i have
for example um four uh custom entry points there for example an entry point in germany and france
what will happen is for example you start an application the application will check if there
is a preferred route if there is a preferred route it will try it will instruct the clients
to connect to the to the first gateway in that in that list and in this case this will be germany
if this is successful then um it it will continue to use the the traffic over over that gateway
if it would not be successful it will check if there is a secondary road
it will instruct the gateway to connect the other secondary road if it's not successful
it will try the third one it will try the fourth one and uh at the end if none of those
routes would work it would fall back to the to the default uh route that was used for example
to to log in to uh to parallels russia so you cannot start an application if you have not
been logged in so there is at least one road which works it doesn't have to be one of the roads and
in the preferred road because at least the the road that was used to do for example
so then in in the last uh last case it will fall back to the to the original uh
connection and and and use that one to start the application so preferred routing could also be
like one of those features to to cope with like more complex data center scenarios where you have
like data center a data center b and you you don't want for example clients or or data to go from
data center a to data center b to go to the outside road you want to keep all your data
in the same data center then the preferred road is one of these features that can be
another enterprise feature is what we call multi-hop gateways so what is this about
some enterprise customers for example want to terminate all their traffic first in their dmz
and then as a matter of speaking with a single firewall port go to the internal internal networks
it is technically possible also to set up gateways in what we call forward mode so
if you if you have for example two gateways you can go to your first gateway and change the mode
from from normal mode into into forwarding mode and then you could for example put that gateway
in your dmz network and then forward from your gateway in your dmz network to your gateway in
your internal network only open that one firewall port and then it's the the the private secure
gateway in the internal network that then will for example speak to the connection broker it's
also that one that for example will host the the web portal so all the all the all the logic is
then done on that on that gateway and that connection broker inside and and the first
gateway which which is in the multi-hop scenario is pure there to terminate the traffic and to
to to to look at inside the network also for these more complex scenarios there is no need for any
extra licenses or any extra management tools everything can be managed from the
from the centralized management portal. The last topic I want to quickly touch before
I go to the the Q&A part this is on the on the client side so from the RAS console
there are several ways that you could for example also pre-configure the clients
we see this very often in enterprise context that you want to allow users to once once the
software is on to the device that they don't need to configure themselves the connection settings or
or the or the or the client settings and we can via email or via dns text records do some
some centralized management from the from the RAS console but it's also technically possible to
for example use like a third-party software distribution tool like Intune to install a
parallel client with pre-configured pre-configured settings and if you take a look there are two
kinds of settings we have what we call client settings is the clipboard allowed do we allow
printer redirection do we allow do we allow access to the local local devices and this is also for
example typically settings that you would configure with a policy and I'm not covering
that today in detail but the client settings you could have like for example a first setting
on the on the device and then once the user is logged in that as of then he will automatically
download a policy and then the other policies you can keep them up to date or out of them
and then there are also connection settings that's for example like connect to production
connect to non-production connect to customer one so the connection and if you want to
pre-configure the client settings you can run the client you can make your own configuration
and then you can export your settings into like a 2xc file and then there are some there are some
parameters that you can specify to msrx when installing the client to use that pre-configured
configuration file or for using it or if you call the file 2x settings or 2xc and you put that in
the same folder as your as your client msi it will automatically also take these these settings
so that's that's one way to to pre-configure clients the other thing that could be done and
that's the most specific towards those connection settings and this is more important for example
if you're talking if you're talking about like shared shared devices shared devices or computers
where multiple users log in typically you don't want to store people's their credentials onto
these shared devices and maybe this is a device in a hospital where where the morning shift has
ended and the second shift starts then you're using the same device but you don't want to use
of course the same same credentials and there is a possibility that once you have made your
your configuration that you're exporting your your your connection settings and you can do
that again with msrx and then you're using the shared device parameter and then there is a little
trick and i don't have a lot of time to go into details but you can specify 0 1 or or 2 if you
specify 0 it will automatically wipe everything if you specify 1 it will only export the connection
settings but not the username and the passwords and if you do set it to 2 it will also store the
connection details so including the the username and the and the password and then once you have
that file you can do of course the opposite you can then for example automatically deploy a rask
client with the shared device setting and then for example say like one import and your file
and then it will import those configuration settings without importing the username and the
password if you want to see it in action there is a youtube video on our techbytes i'm showing you
the techbytes also uh later on but but this is a also like a nice thing which could be super
interesting in an enterprise context on how to pre-consider your your device settings and then
of course last but not least you can of course also combine the the client settings with the
shared device setting so you can combine these parameters all all together so so that's also
possible and which brings me to the end of of this of this technical webinar
this is actually one of the first times we do these kinds of pretty advanced technical
webinars and so let me check if there are any questions let me open the q a part and let's
quickly look if there are any questions that came in if there would be any questions don't be
don't be shy just just ask so there is a there is a question and the question is if you have
two farms does each farm consume different licenses we are and i cannot read the rest of the
question and so yes so if you have two farms you will always have two licenses of course the the
parallels licensing is based on concurrent users so i cannot read the full the full question but
if you're migrating from one farm to another let's assume that the users are not logged in at the at
the same time so they will typically not not come twice at the same the same moment i don't know
your specific specific scenario but feel free to feel free to reach out we can get into contact and
see how we can we can help we can help each other let me see there is a question we are not yet a
customer but we're planning on trying parallels in 2026 thank you for for that so my question
is it possible to expand a single farm single site to a single farm multi-site yes that's
perfectly possible so you can add or delete sites at the at any time there is a there is a that's
not not an issue you can start actually when you when you deploy your your your primary connection
broker it will automatically set up as a as a single site single farm and then you can add a
you might for example then delete your secondary site because you don't need it anymore so
it is technically possible to to to alter from a from a single single site to a multiple site
um let me double check if there are any other questions i don't see any new questions coming
um i would say don't uh don't be shy if there are like any uh any questions uh please feel uh
feel free to uh to uh to let us know reach out um there is my email address um on the on the
slide so if if you prefer to ask at the private just send it to me uh i'll either try to reply
myself or or get you into contact with the any of our um colleagues that that could help you with
with this but uh with uh with that said let me quickly check the last time if any new questions
came in i don't believe so so with with uh with no new uh questions in and almost at the at the
time i see i've seen that there was a lot of people in this webinar so i want to thank all
of you for for joining and if you if you like the webinar uh let us let us know if you have
any suggestions for other webinars into this kind of style let us know we're considering webinars
also on for example single sign-on and some of the other more more advanced functionalities into us
so please let us know tell us what we did well tell us what we can do better and for those of
you who are at the end of your day have a nice evening for those of you joining from uh us or
more west of me have a nice rest of the day and thank you for for being here
