The release of XenApp/XenDesktop version 7.12 introduced couple of new FMA services (primarily used by LHC) — time for an update. As you might be aware, I have written multiple articles on the FlexCast Management Architecture in the past (including my book) talking about its core services, their responsibilities, capabilities, communication channels/interfaces and so on. Throughout the past two years I also came up with a nice graphical overview (at least I like to think so) representing a Delivery Controller including all main FMA services. This article/post is meant to provide you with a continues update on the FMA and its primary core services, graphical overview included. Each time something changes you’ll read about it here – as soon as NDA has been lifted of course.
I’ll start with a short introduction before highlighting each of the services individually – most of the below sections, except for the NEW additions can be found in my book as well (slightly altered though). I’ll make sure to point out what has been changed, added, removed etc. as you go through the list of services.
Internal (Service) communication
Although FMA services run completely isolated from each other, internal communications between the different services takes place using so-called WCF (Windows Communication Foundation) end points (also referred to as service interfaces) over port 80 by default. Though, here I would like to note that port 80 can be changed into any port number you might prefer and that encryption is supported as well.
The endpoint address is represented by the EndpointAddress class, which contains a Uniform Resource Identifier (URI) that represents the address of the service (a secure identity including a collection of optional Headers). Each service has the ability to interact with all other services in order to carry out important tasks and actions throughout the FMA. As a best practice, you will deploy two or more Delivery Controllers, avoiding a single point of failure and providing scalability where and when needed.
As we will shortly see, there are four services that have a special place within the FMA and take on a more prominent role than the others (a.k.a. core services). These are the Broker, Configuration, Delegated Administration and the Configuration Logging service.
FMA fact: While all services closely interact with and depend on each other, at the same time they are also completely separated from each other. Each service is configured to communicate to the Central Site database using its own individual DB connection string. This way, if one service fails it will not affect any the other services.
One of the biggest differences between the IMA and FMA is that with the FMA the Delivery Controller is only responsible for managing and brokering connections to managed as well as unmanaged VDAs installed on your server and desktop machines. It doesn’t host any sessions of its own.
Also, all the ICA / HDX bits, bytes and related services (actually controlling the HDX functionality throughout a session) reside on the VDA itself, again relieving the Delivery Controller from any additional tasks. Because of this approach, it is also easier to maintain the code and different operating systems (server and desktop) can be deployed within the same site.
FMA fact: Keep in mind that if you change something for one specific service, like the DB connection string, for example you must do this for all the other FMA services as well.
Let’s go over each of the services one at the time. Where applicable I’ll elaborate a bit more on the inner workings, any special considerations and/or relations to other services like the XML / STA Service, for example.
FMA fact: All FMA services run under the NT AUTHORITY\Network service account. Also, when authenticating to the Central Site database (this is where the Configuration Service plays an important role as well) all services use the local computer account of the machine that they are currently running on.
1. Broker Service
The broker (XML) Service is probably the best-known one. By the way, it’s indicated as Principal and Secondary (in green) on the image above, with the Synchronisation Service in between. From a Delivery Controller point of view, it brokers new and manages existing sessions, handles resource enumeration, the creation and verification of STA tickets, user validation, disconnected sessions, and more (see the list under 1.1 as well). From a VDA point of view it takes care of all communication to and from the Delivery Controller. It does this by communicating with the VDA Citrix Desktop Service, a.k.a. BrokerAgent.exe, which is part of the desktop as well as server VDA.
Note that the STA (Service) is also part of the Broker Service, and has been as of Presentation Server 4.0. Before that it was written as an ISAPI extension for Microsoft Internet Information Services, or IIS. As of XenDesktop 4.x the XML service (ctxxmlss.exe) has been rewritten in .NET and became part of the Broker Service as well. So, the Broker Service is built up of three separate services (or four if you also count in the newly introduced Principal Broker Service, see below) all handling different tasks: it brokers connections, it enumerates resources, it can take care of authentication (token) and it acts as the Secure Ticket Authority, generating and validating STA tickets; however, this only applies to resources launched through a NetScaler, and last but not least, as of version 7.12 it is also involved in managing the Local Host Cache.
NEW – As of XenApp/XenDesktop version 7.12 the Local Host Cache (LHC) got re-introduced (though it’s new to the FMA). One of the main LHC components is the Broker Service, however, when the LHC is involved it is also referred to as the Principal Broker Service (as you will find out shortly there is also a High Availability Service a.k.a. the Secondary Broker Service). Think of it as a sub-service, just like the XML Service, for example. The Principal Broker Service will accept connection requests from StoreFront and it communicates with the Central Site Database just like before — brokering connections, taking care of load balancing and so on, while it also (continuously) interacts with the Configuration Synchronisation Service as well as the High Availability Service when the LHC becomes active. See points 2 and 3 for some more detailed information.
As mentioned, the Broker Service has some huge responsibilities within the FMA. As such, my FMA partner in crime, Mister Mick Glover a.k.a. XDtipster on Twitter, often refers to the Broker Service as being the workhorse of the FMA. Besides some of the tasks already highlighted, one of its most important tasks is the registration of all VDAs, including ongoing management from a Delivery Controller perspective. To give you a complete overview of the main tasks and responsibilities of the Broker Service, have a look below.
- As already mentioned it takes care of VDA registration, resource allocation, connection brokering, licensing enforcement, amongst other tasks.
- During the initial user logon process it will validate the end-user’s identity, based on credentials received through StoreFront.
- Based on your configuration, it can take care of XML based user authentication (token creation).
- Functions as the Principal Broker Service when LHC is enabled.
- It continuously interacts with Configuration Synchronisation Service.
- It is involved in HDX policy management.
- It manages the overall power state of desktops and server machines when run virtually, starting and stopping VMs based on usage and on administrator configuration, including idle pool management.
- It temporarily stores user credentials to allow users to be logged into virtual desktops without having to re-enter credentials (single sign-on).
- It keeps track of virtual desktop state, based on information received from virtual desktops. As such, it will take appropriate action when needed.
- It will participate in the initial load-balancing process, deciding which desktop or server to connect to. This information will eventually end up in an ICA launch file.
- Administrators will use the Broker Service, although they may not actually know it to log off sessions, define Machine Catalogs and publish virtual desktops based on computer identity.
- It exposesHypervisor state and alert information.
- It will also handle all power management features, including but not limited to: power policy rules, reboot schedules and cycles, pool/buffer size management, and remote PC wake on LAN.
The Broker Service is somewhat special in that it also houses multiple s0-called Site Services, let me explain. Site services provide Site-wide maintenance and housekeeping functionalities within and a XenDesktop Site. They take care of things like managing connections to your Host Connections, checking up on session idle times, managing reboot schedules, cache maintenance (refresh) and more.
Before I go any further you need to know that there are eighteen Site services in total, with each having its own responsibility, or multiple in some cases and that they are part the Broker Service. More specifically, each individual Site service will only run on one of your Delivery Controllers (within a Broker Service), creating a distributed model. As soon as a Delivery Controller misses a *heartbeat with the Central Site database, all Site services running on that particular Delivery Controller will be transferred to one of the other active and still considered healthy Delivery Controllers. Again, they will be moved from one Broker Service to another.
*A heartbeat message is exchanged between a Delivery Controller and the database every 20 seconds with a default timeout of 40 seconds.
FMA fact: While it is considered a best practice to keep all Delivery Controllers equally configured, Site services are the exception to the rule, so to speak.
At runtime, when a Delivery Controller becomes active the Site services will automatically be divided between all active Delivery Controllers within your Site. Do note that although you as an Administrator have the ability to assign certain Site services to a particular Delivery Controller if you want or need to – this is not a recommended or supported approach. The election mechanism is controlled by the contents of the Central Site database, the FMA will take care of this for you.
The eighteen Site services are:
- Hypervisor connection.
2. NEW – the Citrix High Availability Service a.k.a. Secondary Broker Service (LHC)
I already briefly mentioned the Secondary Broker Service/High Availability Service while referring to the Principal Broker Service a couple of paragraphs back. Together with the Configuration Synchronisation Service (see below) all three services reside on every Delivery Controller in your environment — assuming you are running version 7.12 or later.
When an outage occurs, the Principal Broker Service will no longer be able to communicate with the Central Site Database, as a result it will stop listening for any incoming StoreFront and/or VDA information and it will instruct the High Availability Service/Secondary Broker Service to start listening for incoming connection requests and handle them accordingly.
As soon as a VDA communicates with the High Availability Service/Secondary Broker Service a VDA re-registration will be triggered. This way the High Availability Service/Secondary Broker Service will receive the most current session information related to that specific VDA (who is connected to which machine, for example). In the meantime, while the High Availability Service/Secondary Broker Service is handling new and existing connections/sessions, the (Principal) Broker Service will continue to monitor the connection to the Central Site Database.
As soon as it notices that the connection to the Central Site Database has been restored it will instruct the High Availability Service/Secondary Broker Service to stop listening for, and handle new and existing connections/sessions. From this point on it will resume brokering operations as before, basically repeating the abovementioned steps of VDA registration to get up to speed with the latest connection/session information.
Finally, the High Availability Service/Secondary Broker Service will remove any remaining VDA registrations and will again continue to update the local SQL Express database (together with the help of the CSS) with any configuration changes from that point on, as highlighted before.
3. NEW – Configuration Synchronisation Service (LHC)
Every two minutes the Principal Broker Service will be checked for configuration changes. If a configuration change has been detected it will be copied over, or synchronised to the High Availability Service/Secondary Broker Service. This is the primary task of the Configuration Synchronisation Service (CSS). These configuration changes include but are not limited to published icons, changes to Delivery Groups and Catalogs, certain Citrix policies and so on. It will not include information about who is connected to which server (Load Balancing), using what application (s) etc. this is referred to as the current state of the Site/Farm, which is not considered a configuration change.
All (synchronised) data is stored in a Microsoft SQL Server Express LocalDB database (notice the LocalDB part, this is different from SQL Express), which resides on the same Delivery Controller. In fact, each time new information is copied over, the database will be re-created entirely. This way the CSS can, and will ensure that all configuration data stored in the Central Site Database will match that of the data stored in the local SQL Express database keeping the LHC current.
As a second prime responsibility, the Configuration Synchroniser Service will provide the High Availability Service/Secondary Broker Service (s) with information on all other controllers within your Site (Primary Zone), including any additional Zones you might have configured. This way each High Availability Service/Secondary Broker Service will know about all the other available High Availability Services/Secondary Broker Services within your (entire) Site.
Communication between the various High Availability Services/Secondary Broker Services takes place over a separate channel based on an alphabetical list containing the FQDN’s of the machines they (the services) currently run. This information is used to elect which High Availability Service/Secondary Broker Service (read: Delivery Controller) will take over within that specific Zone when the LHC becomes active because of a DB failure or another outage of some sort.
4. Configuration service
All FMA services need to register with the Configuration Service on start-up so that it knows they are all good to go. This is one of the main reasons why the Configuration Service has such a prominent role: it handles all inter-service communication within the FMA. Again, I would like to quote Mr Glover on this one: the Configuration Service is the glue holding the FMA together.
Located at the centre of the FlexCast Management Architecture, it holds and manages a list of all FMA services, allowing them to advertise their WCF addresses, or end points (service interfaces), including the functionality that they provide. Only after a service successfully registers with the Configuration Service, when adding in more Controllers, after a reboot or during Site creation, for example, will it become active and able to communicate with other FMA services.
Once all services have successfully registered themselves, the Configuration Service will share a listing of all active and registered services as being active Site members, including their main responsibilities, or capabilities and service (communication) interfaces.
As soon as an individual FMA service needs to communicate with one of the other FMA services it will first (need to) contact the Configuration Service to get a copy of the services listing mentioned earlier.
After an FMA service successfully queries the Configuration Service and the listing has been received, this information will be cached for five minutes. This is mainly to ensure that the system isn’t being overwhelmed with service listing requests, preventing the Configuration Service from becoming a potential bottleneck. At this point the requesting service knows where to find the other services, what they are capable of (responsibilities) and how to communicate with them (end points/service interfaces).
As a side note, the Machine Creation Service and the Machine Identity Service both communicate through the Host Service to find out about the configuration and connections of the underlying Hypervisor/Cloud connection (Host Connections) if any, including the storage and network configurations needed for virtual machine provisioning. This information will be cached for one minute as opposed to the five minutes mentioned earlier.
FMA fact: Each FMA service can query the Configuration Service to look up other services using the listing mentioned earlier. In short, service registration and communication are both reliant on the Configuration Service. It will also store configuration metadata for all services, relieving Active Directory.
The Configuration Service directory stores the Active Directory machine account identifier (SID) for each service that has successfully registered with it. At the same time, this information will also be stored in the Central Site database where it will be accessible to all Delivery Controllers including the services that they host. Only when the machine SID of the accompanying FMA service is listed, and known by the Configuration Service will communication between FMA services be possible.
When a service with an unregistered machine account contacts the Configuration Service for the services listing it will receive an access denied. The only exception to this is the ‘Network service’ account: it is always allowed. Viewing and validation the successful registration of FMA services can be done through the PowerShell SDK. Use the following syntax:
Get-ConfigRegisteredServiceInstance -InterfaceType sdk | select serviceaccount, interfacetype, servicetype | format-table
FMA fact: If you would like to refresh the cache of one of the FMA services (remember the five minutes), all you have to do is restart the accompanying Windows Service. The cache (services listing) is retrieved during service start-up.
5. Configuration Logging Service
Monitors and logs all configuration changes made within a XenDesktop Site, including all Administrator activity. Depending on its configuration, no Site changes are possible when its database is unreachable, making it one of the four core services as mentioned earlier. The data itself can be stored within the Central Site Database or a separate database can be created, which would be the recommended approach. As of XenDesktop version 7.7 a separate location / database can be selected during the initial installation configuration process.
6. Delegated Administration Service
The Delegated Administration Service is also considered to be one of the FMA’s more critical services. All other FMA services will need to communicate with the Delegated Administration Service in order to validate if they have all the proper permissions and/or rights needed to make the necessary changes to the Central Site Database. Next to this, it manages the configuration and administration of all delegated administrative permissions. Thus, if this service becomes unresponsive or unavailable site-wide configuration changes will not be possible.
7. AD Identity Service
Handles all Active Directory computer accounts (identities) related to XenApp / XenDesktop virtual and physical machines.
8. Machine Creation Services
Handles the creation of new virtual machines. When this service is unavailable no additional virtual machines can be created, at least not using MCS. Also note that MCS is only capable of provisioning/creating virtual machines, not physical. If you want to be able to ‘service’ physical machines you will need to use Provisioning Services.
FMA fact: If you do not configure a Host Connection within Studio, when creating a new Device Catalog, the option to use MCS as a provisioning mechanism will not be available (greyed out).
9. Host Service
Manages all connections between the physical hosts, the Delivery Controllers and the underlying Hypervisor (s) or Cloud platform (s), both referred to as Host Connections. As far as the Hypervisor goes this can be either vSphere, XenServer, Hyper-V or the Nutanix Acropolis Hypervisor. This is where your virtual server and/or desktop VMs live. Physical machines are still optional as well, but again you’ll use PVS instead of MCS. The following Cloud platforms are supported: Amazon Web Service, Microsoft Azure and CloudPlatform. As highlighted earlier, the Host Service is also responsible for discovering and managing the connections and configurations of the Hypervisor, network and storage that are required and used by machine provisioning operations. It does the same for any cloud platforms/connections you may have set up.
9.1 Hypervisor Communications Library
The HCL is used by several FMA services, the Broker, Host and MCS to be a bit more precise to provide an abstract Application Programming Interface, or API for interacting with the underlying Host Connection, or multiple. This will ensure a consistent and consequent representation of the configured Host Connection including network and storage resources. As an example, while multiple Hypervisors are supported this also adds to the complexity of the code, this is where the HCL comes in, it functions as an abstraction layer. As a result, when a new version of a Hypervisor is released or an existing one needs to altered, Citrix can quickly add support without the need to replace the code in multiple places. Again, the same applies to Cloud platforms as well.
10. Environment Test Service
Takes care of all Site-wide tests, initiated from Studio. You can run tests on your Delivery Groups, Machine Catalogs or even on your entire Site configuration, and more. New tests are introduced with every new version of XenDesktop/XenApp going forward.
11. Monitor Service
Monitors the overall FMA architecture and produces alerts and warnings when it finds something potentially wrong. These will pop up in Studio or Director. Note, however, that, although these alerts will tell us that something is potentially wrong, they won’t tell us what is wrong or where to look for answers. Therefore, FMA services are best checked/monitored using PowerShell code typed in directly from a PowerShell Command Prompt.
For example, try using the Get-BrokerServiceStatus and/or Get-ConfigServiceStatus cmdlets to view the status of the Broker and Configuration service. There is a ‘Get-‘ command for each FMA service.
12. StoreFront Service
This takes care of your StoreFront deployment, which can be added as well as managed directly from Studio. Note that your StoreFront can and probably will appear twice within Studio. Once under the console root as an integral part of Studio (it will be there by default) which is needed to be able to configure the Citrix Receiver as part of your published hosted shared desktop (s) images directly from a Delivery Group. And a second time (as a separate node) enabling you to manage your StoreFront deployment as if you were using the separate MMC based StoreFront management console. The latter gives you the option (s) to change its look and feel, authentication methods, adding and removing Stores, and more.
13. Analytics Service
The Analytics Service does as the name implies: it collects analytical data used by Director/Studio (custom reports, to name one). It is also leveraged by the Citrix Customer Experience Improvement Program (CEIP), which will be enabled by default — this applies to the Citrix Call Home functionality as well by the way. All data will be shared anonymously, encrypted and, as always, will be used for the greater good.
More to come
As you might have noticed on the graphical overview, more services are on their way. Unfortunately I can’t share any additional information as all this is still under NDA. Just know that the services are already there, just not activated, usable and/or viewable. I will make sure to update this post as soon as something changes. Thanks for reading.
Again, I would like to thank Martin Zugec @ for his assistance throughout this article and other articles I have been working on, some already published, some still to come. To be continued…