Citrix Synergy 2015 Day 1
- Published: Wednesday, 13 May 2015
From 12 to 14 May Citrix Synergy 2015 was held in Orlando. In this blog I will describe my experiences of the sessions that I attended. On this first day I attended four sessions.
The first session I attended was provided by Aaron Parker, Helge Klein and Shawn Bass titled I’ve got 99 problems and folder redirection is every one of them. The session started that all three of them explaining their experiences of Folder Redirection, followed why folder redirection is being used: Fast logons, abstract user data. Folder Redirection is always a tradeoff, be careful resource contention impacts the user experience. In this session they now compared SMB 2.1 versus SMB 3.02, based on client Windows 8.1 or Windows Server 2012R2 and the file server is Window Server 2012R2 or 2008R2 and discussing the results. It’s difficult to blog the results out of graphs so get the slide deck for all information. Remarkable is de CPU usage on a file server when the disk systems is used a lot for HDD, while SSD do not have behavior. Helge referred to test executed by Magnar Johnson about the results of different SMB version connect to each other. Helge dived deeper in this part and shares his results which more or less the same for the different versions (so the results are different than Magnars). Nice tip to define which version of SMB is used via PowerShell. Aaron continued with some tips and tricks: simulate client load to determine file server breaking point, watch overcommit of the hyperviso, multiple vCPUs, scale out File Servers, size for peaks and expect performance of anti virus, followed by Folder Redirections alternatives like Work Folders, AppSense DataNow, Citrix ShareFile and User Profile Disk. They ended with summart: no troughput difference between SMB 2.1 and 3.0-, lower average CPU usage with SMB 3, anti virus has a big impact, size file servers for peak load, don’t overcommit resources, legacy IT needs to adopt to modern mobile foldes.
Next session I attended was the Workspace Pod provided by Momchil Michailov and Daniel Feller. They started by the amount of IOPS that were mentioned during the keynote. They started with discussing the advantages and challenges between a decentralized and centralized infrastructure nowadays. The future according Daniel is using the WorkSpace Pod on the branch office managed via the Citrix Workspace Cloud. Another scenario is high density XenApp. He started with discussing virtualization overhead on CPUs, PSV possibilities on physical machines and availability on VMs. This leaded to HP MoonShot, followed by adding Melio into the picture. Momchil continued describing the WorkSpacePod, which can be converged or hyper-converged. It’s also including the data. He continued with describing features of the WorkspacePod like LTA (storage and performance module enhancement), SnapShots and more. The demoed with two sites, were a PowerPoint is distributed between two Pods.
Third session was ShareFile Enterprise integration and field tips. They started explaining shortly the possibilities of ShareFile. They dived more into detail of the components. Anton started with the StorageZones and the connectors to the storage zone. Next they discussed authentication (ShareFile only, SAML via ADFS, SAML via XenMobile, SAML via NetScaler). Topics were User Management Tool, the logon end-user possibilities, NetScaler integration (Jarian stated that the wizard for ShareFile in NetScaler should be used) and XenMobile integration. Next topic was troubleshooting. Jarian first tip was to remove the NetScaler out of the configuration. Anton mentioned log files of the StorageZone controller (based on IIS). Check firewall/proxy configurations. Time sync is important with XenMobile, re-save your ShareFile settings. Last topic was ShareFile API’s. Jarian mentioning that PowerShell is the easiest, Visual Studio and JSON are more difficult. Jarian also explained the basic steps using the SDK.
Originally I would have presented my session at the end of this day, however the exhibition hall was closed during this timeframe (the session would be in the expo hall). So I was rescheduled to another day. This made room to attend another session. I selected XenDesktop 7.6 Architecture dealing with failure by Thomas Gamull. The first part was very entertaining, explaining about failures and what currently is done about (like NetFlix Chaos Monkey). According to Thomas automation is key, focus on reducing cycle times (what is wrong+how to fix it+implement fix), immutable servers (servers are rebuilt from scratch for changes), what user can handle. Thomas don’t agree with Aaron as he likes Redirect Folders (except appdata and desktop). Profiles should be as light as possible, use mandatory as possible are other point Thomas advices. Profile replication can only be active/passive (one way replication). Thomas advises to use SQL mirroring for the databases and use local disk with PVS. For network Thomas advises NL GSLB for multiple sites (active/passive is easiest to setup), all components should be load balanced. Next Thomas showed one of his architectures.