Wilco van Bragt - LinkeIn Wilco van Bragt - Twitter rssa 

Citrix Technology Information Exchange 2011

On the 23rd of June 2011 Citrix Benelux organized for the first time (as far as I know) the Citrix Technology Information Exchange (CiTIE), which was a technical event on invitation only basis. I personally like the event a lot, there were good sessions and I really knew a lot of people, so many to talk and discuss during the day. The day exists of 4 slots of three parallel sessions. In this blog you will find my experiences of that day.

The location (the asument park De Efteling) is pretty close to my hometown so I was on time and the day started with lots of saying hello to people. It was good to see all those guys again. There was no real official opening or a plenary keynote, so directly starting with the parallel sessions.

The first session I attended was XenDesktop.Next the future of on-demand desktops and apps by Paul Murray. After some basic introduction of VDI the session started with a recap of current features available in XenDesktop 5. The first topic was the simplified deployment describing the desktop suite, the quick deploy option and the desktop deploy option (including the differences between those two). Also the catalog (collection or pool of similar machines types populating desktop groups on one or more hypervisors) was discussed. Paul continued with the Machine Creation Services (MCS). According to Paul Citrix tested MCS in the same setup as Citrix Provisioning Services (CPS) receiving the same results of scalability (although it attacks more on the storage as CPS do because CPS can do the write cache on other locations, while MCS does this always on disk). This part lead to some questions and discussion why Citrix started MCS and what this means for CPS according to support and development on the product. Paul continued with explaining the MCS process (copy of golden image to master vm image, admin provisions amount of VMs, creating ID disk and difference disk for each vm so cache is on disk/storage, OS reads from base disk while OS writes go to difference disk.

The session continued with upcoming features in XenDesktop in the XenDesktop.Next, like HDX Nitro Technologies, Adapative orchestration (determine where rendering should take place inspecting client, network and server), HDX Mediastream, Aero Reduction, Scanner support over WAN, Client Drive mappings enhancements (support UNC path, faster performance), webcam support down to 300-600kbs and high latency, flexible QoS (using Citrix Multistream) and Desktop Directory enhancements (more HDX info, XenApp support, Highlight problem areas).

Next was describing the Intellicache feature on XenServer. Intellicache is creating a caching object on local storage on XenServer (reducing IOPS) for reads. There is one read caching object per XenServer, reads not available in the cache will be read from the image (on storage). For each VM there will be a cache file for write (for the difference disk). Also this part created some good questions and a good discussion again on the possibility to create a shared write cache reducing more IOPS. The last topic was where the future goes. Summarized Citrix is working on integrating XenDesktop and XenApp into the same components in one product.

The second session I attended was by Walter Hofstetter with the title XenClientInDepth. The session started introduction of XenClient (hypervisor 1, simultaneously OS, hardware idenpendent VMs) and the synchronizer (distributing and update). The next topic was the XenClient XT, the hardened client hypervisor with features like hardware assisted security, network isolation, service Vm's). Walter described the architecture of both versions showing the differences. The additional security has also some downsides like no support for the synchronizer and more memory usage.

Walter explained in detail which components were used and where they were based on like Midori, DBUS, V4V, open Emmed, Xen 3.4.4 and so on. This was for me a bit too much information as I don't know much about Linux. 

The session continued describing the VM life cycle describing the image creation, uploading to the synchronizer,   the two available deployment types; Static (one disk for one person) and Dynamic (splitting up in three disks; system, user and app) and the update possibilities and process behind it. Walter showed a nice chart with the characteristics of both types.  Walter also showed the disks created on the system during deployment, enrollment, updating and creating back-ups.  The last topic was What's New in XenClient 2: removed requirement VTD, AMD support, more hardware vendors (Fujitsu, Toshiba, Panasonic and NEC, Over the air upgrade of the XenClient software, 8 GB per VM, Win 7 SP1 support, Ubuntu support, improved Windows 7 audio support, cached VHD Downloads (if image is locally available on DVD/USB and will be used for deploying instead of the network), scalability and backup optimizations.

After lunch I visited the session Migrate your organization to the Citrix Receiver the easy was by Simon Frost. The session stared according to Simon with fluffy sheets to get the point of the topic. Topics passed like Appstores, IT consumeration and so on. Then Simon started with describing the Reiceiver vision Citrix has (standard client of each platform, single authentication, integration GoTo products) and the required components (Receiver, Delivery Service and Applications). When the platform has no specific client, Citrix is developing a HTML5 based client. This client arranges true clientless access, automated fallback, first supported on Chrome OS, but the client will have less features then the platform Receivers. Simon continued with describing and explaining the architecture behind the Delivery Service (middle tier between provider layout and receiver) of many services like desktop, apps, Saas (Citrix Cloud Gateway), data (sync data on all receivers on encrypted local storage) and collaboration (Goto integration). It was again pretty big dive into the architecture describing the available services and what possible extensions could be (both Citrix as 3rd parties) and the way the authentication mechanism works (federation through security tokens). Next topic was the upcoming Web Receiver (the successor of Web Interface with built-in self-service) including a small demo. Simon also stated that on WI no new development will take place anymore. Last topic was pointing to the merchandising server (virtual appliance) for deploying the receiver.

The fourth and last session of the day was XenServer: Beyond the management console by Michael Mellor. The session started with the general architecture of XenServer (linuxparavirtualization, intelVT, Controller Domain, Linux Bridge v Open vSwitch). Next the difference network object were discussed (PIF, Network, VIF, Bond), design considerations (VLAN, ACLs, Bonding, Netflow statistics) and how those can create using the command line. The same steps were performed for the storage components (VDI, PBD, VLM, SR, VBD), design considerations (type of storage, thin provisioning/snapshotting required, back-up requirements, performance) and the CLI command to create those objects. Michael also mentioned and showed some tools like XE, XenTop (CLI based semi GUI viewing current VM performance and XSConsole (GUI in the CLI environment), followed by some tips and tricks (CLI XE command for bypassing full blown socket appearance in VM for licensing issues, log file locations and creating graphs using RRD. The last topic was Project Boston, the upcoming XenServer version with features like product simplication (single ISO installation, full integration storage links and site recovery), enhanced architecture (Xen 4.1.0, open vSwitch default, SR-IOV improvements), XenDesktop enhancements (GPU pass-through full support, one card per vm), guest OS updates, Self Service/Cloud Building Tools, MSSC integration and NFS for HA support.

In total it was a good event with mostly good content. I hope more of these events will follow in the future.