VMworld Session Notes: Impact on the Delivery of Healthcare IT Services

August 31st, 2010  / Author: admin

Below are the cliff notes for VMworld session EA7829: Virtualization Impact on the Delivery of Healthcare IT Services

- Though group discussions they found that printing is a problem. Printing has also been a dirty word with Healthcare. Proof you need to be in the field to get IT.

- Imprivata has leading login solution for medical staff. Integrates a web can to provide an extra layer of security.

- James Fitzgerald committed that Healthcare was viewed as cost center. Cost savings can be used to reposition staff to helping doctors on the floor.

- Lots of buzz words of flexibility, agility, compliance.

- Virtualization is reducing data left on the endpoint.

- IT people are paid less in the medical field and hence the best people are not implementing the technology.

- Your data isn’t secure today even without the cloud. Most errors are human based and proper processes need to be in place.

- Virtualization is allowing community based healthcare IT shops to provide centralized security.

- Scott Dresen from Spectrum Health System commented that major vendors need to be engaged early on and get their buy in to ensure success.

- James Philbin PhD from Johns Hopkins commented that virtualized PACS run better than physical.

- PCoIP has enabled PACS images to be remotely read from anywhere in the world.

- VMware View 4.5 allows full USB support allowing mics and card readers for zero/thin clients

VMworld 2010 Opening Keynote: Red Cross of Virtualization

August 31st, 2010  / Author: admin

Today the IT community gets the retooling to make the cloud buzz a reality. Automation and an open choice are going to be the onramp for the cloud for future thinking companies. As a VMware Health care customer it’s exciting to see the endless benefits that are now available.

In Canada, a health care provider is not going host information in the USA with the Patriots Act, but the private cloud will allow us as Health Regions to work and share information more easily. Our business value as a Health Region should be to have better patient outcomes. Better patient outcomes are possible with more flexible and reliable application delivery.

VMware announced a new construct of a Virtual Datacenter. The Virtual Center has the ability to create and mange applications in a secure Hybrid Cloud. The ability to move on-demand, whole application stacks will have endless benefits. I think the new open frameworks will facilitate EMRs which have been so hard to get off the ground in Canada. EMR’s will be able to be deployed regardless of the end point or the backend infrastructure. In the end, IT will be delivered as a service.

The future will have service catalogs based on organizational business units. The service catalogs will be tied to SLAs that users can demand and change when needed. The backbone to this delivery is vCloud Director. vCloud Director creates vApps and have virtualized both security and the network to overcome the hurdles of the past. Security can now follow a VM based on policy regardless were it travels. A secure, fast, deployable and accessible model will be key factor for Health Care.

Along with vCloud Director, Project Horizon which was announced today will allow for user centric model across all devices and platforms. Single sign on and roaming profiles have been a pipe dream in the past but the demo on stage looked very compelling. The traditional paradigm of installing applications soon will be dead.

They future looks very bright for IT and Healthcare.

Notes from vDs Deep Dive: Managing and Troubleshooting

August 30th, 2010  / Author: admin

This is my first break session for VMworld 2010. The lineup started over a half hour before the session. I will try to list the key take aways from the session. VMworld attendees can download the contact about a week after the event but not everyone is here!

While VMware is committed to the standard switch; however, it is easy to see the vIrtual distributed switch will the foundation for many key products. I took the VMware vCloud Director – Install and Config lab and it was heavy vDs. The ease of automation across the datacenter will definitely benefit all the enterprise plus customers.

Notes from the Session

Concepts:
-Central management plane
- vDs contains virtual ports called DVPorts
- Port binding: three types of binding
Static Binding – default configuration, port bound when vmnic is assigned. Has the best performance and scaling.
- vCenter stores all configuration of the switches, on the host the configuration is stored in cache. the switch is operational independent.
-vMotion: migrates the port state of the vm, normal switch does not
-Traceable VM to port migration

Features of 4.1

- Network I/O control: Able to manage traffic flows, iSCSI, FT, VMotion, NFS, vcenter management traffic. Traffic prioritization and Network Resource Pools to form QofS. You can also set bandwidth limits.
- Load Based Teaming: recommend to be used with Network I/O
- Scalability was increased

vDs Management

- Monitor Traffic with port monitoring in the vSphere client
- ESXi, can run ESX-cfg commands form ssh if enabled
- Can use esxtop and resxtop for real-time traffic information, type n once esxtop is running
- Cisco Discovery Protocol is enabled and you can tell which physical ports the vm is plugged into

Troubleshooting Steps
- Check physical nics
- check vDs next, check rx and tx
- check the vm nic
- last place to check physical switch for proper trunking
- use a sniffer to check for anomalies
- try to ping a local vm on a server to test the uplinks
- use esxtop to see dropped packets and retransmits
- verify all nics in a team are connected to the same broadcast.

STP on the access physical port mat cause delay in teaming failover.
MS network load balance should be ran in uni-cast.

This session is scheduled to run again on Wednesday at 1:30 pm. Presenter was Amar Padmanabhan and Wei Zhang.

PCOIP: Taming The Beast On The WAN

August 30th, 2010  / Author: admin

For people in the Virtual Desktop Infrastructure(VDI) space or looking at it will have heard of PC-over-IP (PCoIP). Devopled by Teradici. PCoIP enables remote access to workstations and servers, with the look and feel as if you were sitting in front of the machine. The PCoIP protocol transmits only the changing pixels across a standard IP network to software or hardware based clients. PCoIP is the key technology behind VMware’s View VDI solution.

I think the vast majority will agree that PCoIP on the LAN is superior product. On the LAN no customization is needed for this protocol. Install the View Agent onto your windows virtual machines(VM) and you are off to the races. On the WAN this protocol can be a different story. PCoIP can tell if your network link is congested and will throttle its bandwidth. Your can see this is action with a great article from http://myvirtualcloud.net/?p=751. However, the protocol is still greedy and I don’t think it throttles the bandwidth quick enough. With some work and tuning you can make this protocol work over the WAN for the vast majority of users.

Here are few steps we have done in our deployment of VMware View.

  • Make sure you use SALSA20-256 and deselect AES-128. SALSA20 is faster at encrypting and decrypting the traffic.
  • Load the GPO from View Connection server and adjust the following settings either in Active Directory or locally on your base VM template.
    • PCoIPImageingMinimumImageQuality – The default is 50. The value can be between 30 – 100. I think the default is fine. This value is only going to kick in when the network link is stuffed.
    • PCoIPImageingMaximumIntitialImageQuality – The default is 90. The value can be between 30 – 100. I have had good luck with 70. Tried lower settings but users were complaining about the “Wave”. The screen would ripple as it would build to the Maximum Image quality.
    • PCoIPMaxLinkRate – The maximum session bandwidth in kilobits per second. The default is 0, for no bandwidth constraints. For desktops in rural areas I have set this to 1000 Kbps. With all the combined settings listed in this post most desktops will only use 100Kbps – 200 Kbps. I have lefted it at 1000 Kbps so they have a chance to use the available bandwidth when available. You know never when someone is going visit a flash site. For whatever reason when you set this setting at 2000 Kbps or lower it will better rate limit audio in the PCoIP stream. This was first noticed by Teradici.
    • HKEY_LOCAL_MACHINE\SOFTWARE|\Policies\Teradici\PCoIP\pcoip_admin_defaults\pcoip.maximum_frame_rate – This registry settings will be the one that will help the most in saving bandwidth.
      • Teradici Default = 30 fps
      • HDx Protocol = 24 fps
      • Good Television = 16 fps
      • Office worker = 8 fps

I have tested with 16 fps and the user experience is good for office workers.

With the settings above we have had sites using PCoIP with 30+ active sessions using a 5Mbps link. I hope aids in taming the beast and keeping your users happy. Thanks to Stuart Robinson for all the help and information.

Is My Favorite vSphere Tool Going Away?

August 26th, 2010  / Author: admin

While going through the release notes for vSphere 4.1 I noticed one of my favourite vSphere tools be might be going away. vCenter Update Manager (VUM) 4.1 and its subsequent update releases are the last releases to support scanning and remediation of patches for Windows and Linux guest operating systems.

I think it would be a mistake not to continue on with VUM. The tool can scan and remediate both powered on and powered off virtual machines (VMs). You can easily tell from a single pane of glass which VM’s are compliant or not. With a large virtual desktop Infrastructure (VDI) environment it’s fast and easy to update your templates and linked clones. You can even automatically take a snapshot before you apply the updates in case there is a problem. This tool has shaved off hours on monthly change windows. If I had to do all this work manually I would have to hire extra staff which in today’s market is not going to happen.

Shavlik, the company behind the patch database that VUM relies on, seems to have a good working relationship with VMware. Last year when VMware announced VMware GO, a free web-based service that will allow a customers to set up ESXi, Shavilk was a major partner. It’s hard to believe they would just drop each other but I believe there is a cloud play. Shavlik will be providing cloud-based IT management and patching through their OPsCloud strategy. I believe with VMware’s Redwood around the corner the two companies will offer the proper hooks to each others infrastructure in the form of the appropriate cloud API’s.

It’s likely that the same tools will exist but in different forms and under different names. I am sure we will have another reason to spend the money on another upgrade.