Citrix Synergy 2017 Day 2
- Published: Thursday, 25 May 2017
Day 2 of Citrix Synergy 2017 started pretty early with the first break-out sessions at 8:30u. They planned really good session at the early morning time so the rooms were filled, at least the session I attended. My experiences of this second day are in this blog.
The session I attended was Solving PVS Challenges with a simple shell scripts by Andrew Wood and Jim Moyle. They started off with the usage of PVS based out several surveys like VDIlikeaPro and why you would want to be script (less time, less risk). They continued with the possibilities to build scripts for PVS (MCLI, Soap, PowerShell, PowerShell with objects). Andrew and Jim shared the resources to use to automate your build (Aaron and Tronds MDT articles) and optimization tools like VMware and LoginVSI. Also, the unattended installation of the PVS components. Andrew is also mentioning how you use Smart Tools and are working hard to extend this shortly. Andrew showed a demo of a server installation using smart build to build a PVS server. Next topic Jim was talking about the documentation scripts of Carl Webster (who is doing a tremendous job to build and maintain those scripts). Jim added functionality to the script so it creates an object of the configuration using with Pester. Next topic was Desired State Configuration with a shootout to Iain Brighton. Jim will do some work for PVS with DSC as there is currently nothing available. For creation, the VMs scripts will follow from Andrew and there is script available of Jarian. Jim continued with the automation of replication the vDisks and pointing out scripts Ryan Butler and the Samuel Brewslow with the vDisk Replicator Tool. They summarized the session.
Second session I attended was Keys to a successful XenApp/XenDesktop User Experience by Steve Downs and Shane Smith. The session started explaining why user experience is important (perception is everything, expectation, compromise user experience v user density). Advising for the start of the project were: start the design with the user, repurposing the physical workstation experience and giving users rules to follow. Next topic was logon time (every login to a non-persistent session is first-time login, do not repurpose the thick client logon). Steve touched the Windows 10 Service Branches (Current Branch, Current Branch for Business, Long-Term Service Branch). Citrix only supports CBB and LTSB, also the feature updates were discussed and how they differ. They recommend to uninstall the VDA, upgrade the feature upgrade and reinstall VDA (I think this should be optimized and changed). Steve is referring to Daniel Feller for the optimization on Services, Scheduled Tasks, Default Applications). Steve touched some most important optimization (disable BrancheCache, disable Delivery Optimization, disable Spotlight, consider disable search [luckily Steve discussed it thoroughly, including a suggestion to move the search to the write cache drive. I personally would not disable the search function]. Next topic was Windows Server 2016 (LTSB, No Default Windows Apps, similar services as Windows 10). Steve touches the Citrix Optimizer tool (as an equivalent of the VMware OS Optimization tool). The tools also support templates, which will be released shortly. Shane took over with the design decisions, starting picking the right flexcast model, I liked the table for picking the right one. Next topic was the user profile design, including some best practices (keep profile share close to VDI/XenApp infrastructure, high available config, ensure the use of SMB3, recommended to use DFS-N, no DFS-R active/active). Logically Citrix Profile Management was described. Shane continued with describing Workspace Environment Management. Protocol Tuning was the next topic in this session, with a good shoot out that different releases have different defaults. He recommended the following: actively changing regions à Selective H.264 with ThinWire+, Graphic H.264 and the last one I could not catch during typing the notes). Also, the Enlightened Data Transport (Adaptive transport) including the components required. He continued quickly with Skype for Business with the recommendation the audio quality to set to medium.
Next session I attended Introducing Smart Tools for the Xen product family. They started with a quick overview of the Smart Tools components (Smartbuild à quick Citrix infrastructure build, Smart Migrate à XA65 to XD 7.x support tool, SmartScale à service for AWS/Azure, Smartcheck à troubleshooting/diagnostics/health checks). Next topic was about the usage of Smart Tools based on your license (with select all except Smart Scale, platinum all components). Select also gives App Layering and Workspace Environment Manager. They continued with more details about Smartcheck (proactive health check on prem/cloud, aggregates multiple diagnostics/troubleshooting tools, support XA/XD 7.6 and higher) including sowing the features via a good video demo. Matthew continued explaining how the component actually works, always good to have insights what’s under the hood. Next Matthew shows the custom alert option again via a demo. Actually, you can use this to run your own scripts via Smart Check. Next Carl Webster is on stage to talk about his experiences with Smart Tools based on custom script within Smart Tools. Matthew continued with the roadmap (StoreFront, PVS, NetScaler XenServer support, integration of tools and experts in the product). Jie continued with the demo of Smart Scale showing power down and power on VDAs based on usage including the configuration. Jie continued with some tips and tricks for Smart Scale (use schedule to power on machines, choose appropriate machine size, configure idle and disconnect session timeout policies, configure load index policies in Studio if using load-index as metric, install agent into multiple Delivery Controllers for redundancy).
After lunch, it was time for the third keynote Innovation Super Session by Malcolm Galdwell. The session was opened by Tim Managan with the announcement of the innovation winner award. After the innovation award Malcom came on stage. I did not blog about it as you need the hear the whole session to get the story.
I jumped into the session StoreFront top 10 lesson learned from the field because the keynote continued a bit longer. They were already about scaling in versus scaling out. I could not get there recommendations (at the end it was summarized to scale up not out) however they stated that nog more than 4 servers should be in one server group (six official supported). Next topic was sizing: min 4GB memory. Next topic was the baseURL and changing of the base URL. Receiver & Auto-Discovery can cause issues as Receiver client do not support HTTP redirects (reconfiguration of Store is required). So, have the name right before going into production as clients will still use the old name. Michael continued with XML delegated authentication. With StoreFront, you need a store for each untrusted domains, while this could be done differently with Web Interface. Sarah continued with Gateway together with StoreFront. She showed the flow chart determining the gateway is being used. Name is required to match. She also discussed the optional fields (Vserver Ip should be used if you have more gateways with the same name. Callback is required for smart access or smartcard, otherwise leave it empty). For native receiver, it is required to set a default appliance. Lesson no 5 was about Single URLs are harder than they appear. Sarah discussed the requirements (split DNS, unique gateway callback address, unique OGR address and for native receiver: unique beacon and unique account services address). Nice tip for using Optimal Gateway Routing for using MAS statistics with SSO. Michael took over about multi-site aggregation, which is still continues evolved and StoreFront knows more than you expect for example connection leasing. He continued about Favorites. Subscriptions are stored per server per store and there is a subscription replication that should be configures as close around midnight as possible. Lesson nr 9 is maintain as few stores as possible (exceptions: different authentication methods, differing enumeration/resource). It is possible to hide store (Advertise Store setting). Subscriptions and authentication can be shared between stores (GUI or configuration file). Do not delete the very first store, otherwise a re-installation is required (could be done better). Last lesson is that StoreFront sometimes takes the blame for Receiver (like duplicate icons in Receiver by multiple store configurations for Receiver or Roaming Profiles).
Last session of the day was Independent Citrix Experts: deep dive on Remote Graphics, user experience and GPUs. They started of introducing the team behind TeamRGE. Ruben started off with the statement that Windows is not dead (4-8 million apps). Ruben is stating that also Office, Win10, Browsers and the Adobe Suite leverage from a GPU). Benny continued describing the GPU types (Direct Map, Software Graphics, Shared GPU, GPU Pass-thru, GPU Virtualization). Next topic were GPU manufactures (AMD, Intel, NVidia and the types of GPU they are supporting). Ruben continued with the results of GPU usage out of the survey VDIlikeaPro. Ruben is mentioning the Tech Preview of ADM MxGPU with XenServer, which they used for testing, where .8 version is GPU with 8 GB. Benny describing their set-up used for the testing. Ruben is mentioning the RD Analyzer tool by Bram Wolfs and Barry Schiffer. Wonderfull tool and Benny/Ruben are using the tools for their tests. They continued with the Benchmarking results showing the videos of their tests. The demo shows that packet loss is most important, then latency and last bandwidth capacity.