WinToGo very quick initial thoughts...

I really like the idea of WinToGo so carried out some quick tests which highlighted a few interesting results I thought I’d take the time to quickly share.

Many technical notes explain how to create a two partition USB stick ready for WinToGo, on many machines a single partition stick works just fine, it depends on the BIOS format of the device. Booting from WinToGo does take 60 seconds more than booting from a locally installed Win8, but once booted performance on some USB 2.0 devices was fine, while on other (5 year old) PC USB 2.0 speed wasn’t usable.

The USB 3.0 machines tested have all shown adequate speed but I still need to complete some real world testing around performance.

What I did like was the idea of trying this on a HP thin client T610 2GB device. These devices out of the box attempt to boot from USB first to support flashing the onboard ROM, and have blue USB 3.0 ports on the rear. First attempt booting to WinToGo using the USB 3.0 port failed. So I booted into safe mode and disabled the option to create a page file on the local device internal drive. Next reboot worked, so I suspect the default write cache on the thin client was too small and amended this to 600MB which solved the problem.

What I don’t like is the idea of giving this to users for use on a home PC, which would require the user entering the PC BIOS and changing the boot order!

I hope to carry out some more testing across different PC and load up the full set of applications and enable hibernation mode (shutting down the WinToGo machine isn't very user friendly)

 

 

 

Slow logon time (again)

A common problem which delays logon times concerns IE branding. This issue seems random, effecting all Windows versions running IE 7.0 or IE 8.0. Testing this I found results not consistent, with some clients or servers showing a logon delay, but not a universal issue across all devices. On one system were the below fix was applied logon time was reduced from over 60 seconds to just 4 seconds!

By checking C:\Documents & Settings\%username%\Local Settings\Application Data\Microsoft\Internet Explorer\brndlog.txt shows if the problem exists. This log file records time taken for branding to be processed, from start to finish, this file should take nothing more than 2 to 4 seconds.

Microsoft reference this issuein: http://support.microsoft.com/kb/941158 which details a fix for IE 7.0 MS report the issue does not exist with IE 8.0. (support case open with MS for IE 8.0)

The fix for IE 7.0 requires a new registry entry adding to machines:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Internet Explorer\Main\FeatureControl\FEATURE_PARSING_BRANDING_CMDLINE_FLAGS_KB941158
This key seems to completely disable branding, which I suppose is one why of bypassing the issue, although I’ve never come across a fix that requires a registry key with the KB number in the name, looks like a very rushed job. The file responsible for branding is iedkc32.dll which is located in System32 and SysWOW64.

From a fixed IE 7.0 server, I have copied the above files across to a IE 8.0 server and proved this resolves the issue, although that’s not really a logon term fix!

What size is your page file

Page file size is one of those questions that is often pushed to the bottom of the to do list. For a while with improvements in server hardware, the relevance and impact of this question faded, but with more servers running on x64 loaded with 16 GB or more memory, the question what size page file to use starts to crop up again. Should a server with 16GB memory really have a 4GB or larger page file? Looking just at terminal services, servers running SQL and Exchange have different requirements, it’s easy to see if you need to run a page file at all and if so what size. The simple way is to look at just three values:

Memory\Committed Bytes, (Committed Memory is just the total amount of memory all applications have requested)
Process\Working Set_Total (Working Set_total is what is actually been used)
Page File\% usage (page file usage should be the difference between the two values above)

So for example let’s take a Citrix server hosting 55 concurrent users ..
Memory\Committed Bytes = 6 GB
Process\Working Set_Total = 3.6 GB
Page File = 2 .4 GB
(Committed Memory minus Working Set = page file usage)

From these results we can see a server with 16 GB RAM only requires 6 GB memory to support the applications running, but in this example 2.4 GB is been paged out. So for this server the Page file could be reduce to a few hundred meg just to ensure if a process insists on accessing a page file it can.

G20 meeting - impact on IT

The flexibility of Citrix returns real business value again today. With the G20 summit in London many financial companies have asked employees to work from home or take the day off.

Companies that have utilised Citrix for pandemic or DR planning including a remote SSL VPN gateway, have the flexibility needed to support a large percentage of the work force connecting from home. Looking at one company this morning a fifth of the workforce were already online and connected by 7.30am

Talking to a colleague at one of the largest banks in London, he had provisioned an additional 40 servers into the working from home solution this week providing adequate capacity for 5,000 concurrent users in total , should a London building need to be evacuated.

Finance departments rule

This year finance departments are calling the shots, with a single business aim, reducing cash flow for 2009.
This is a challenge for IT departments but also an opportunity. For companies with existing Citrix or VDI infrastructures some interesting options present themselves.
I’ve seen two areas being exploited. Building upon a well designed existing infrastructure to increase working from home capacity incurs just a modest cost. Yet, its cost savings can be exponential. Enabling more staff to work from home can reduce office space requirements, a major ongoing expense.
The second push, resulting from the present economic climate, is expanding the off shore workforce, again quickly and cost effectively addressed with a Citrx XenApp or VDI systems.

This year doesn't have to be taxing, and for those prepared to adapt and hunt for new opportunities, the next 12 months could prove engaging and profitable.

Speed camera for Citirx

It’s part of the human condition that we take risks. Evident in how difficult we find it to stay at the speed limit. The introduction of speed cameras however, changed behavior, just knowing that we are been monitored reduce speed.

Monitoring in the workplace has existed for years in various forms. Typically used to record employees entering or exiting a building or room, exchanging emails, instant messages or conducting telephone conversions.

Recording a user's Citrix session is akin to introducing a speed camera into your applications. Just the knowledge that a users’ actions are been recorded, makes users behave within acceptable limits, and think twice about any actions they might take.

I've looked at two products that while both record Citrix sessions, are very different, SmartAuditor and ObserveIT.

Citrix with the release of XenApp 4.5 introduced SmartAuditor (SA), which requires platinum licensing. SA records the users’ session in full colour, and also lets you watch the recording a second or so behind real time. ObserveIT, on the other hand, captures a small grayscale picture of the screen after user input has occurred, so each product produces a very different visual record.While Citrix captures the whole session including display changes that might happen without user input and so provides a full recording, ObserverIT stores only changes on the screen generated by the user.

That's the main difference in the recording, but ObserveIT supplement the recording with Metadata about the users actions, for example if a user opened a command prompt this would be tagged, then if the user issues a ping command this again would be tagged, the result: you can quickly scroll down a list of user actions then click to see the related recording.

Both products can be configured to record a server desktop or published application on a per user basis, I’ve used this to only record third party access to applications while not recording internal trusted users.

Session recording is an emerging technology; it will be interesting to see how it matures and is used over the next few years.

Days like today

With the heaviest snow storm for eighteen years, London comes to a stand still. The bus network closes and with tubes and rail severely affected, firms in London start calling their staff to tell them not to bother travelling into work. So, with the estimated cost to the economy predicted to be in the region of £1.2bn, who are the winners and losers?

Well, there might not be any winners, other than the kids in the park! But, you can be sure that the companies who implemented an effective disaster recovery solution were patting themselves on their backs. I’ve worked on a few solutions to support home working via Citrix and SSL VPNs, which have quickly been adopted to become a core component of DR planning.

The dangers of success, however, need to be mitigated. When DR solutions are used, they tend to be pushed to the extreme. So, take care when planning. Typically, the main danger are; Internet bandwidth, ensure the capacity exists to allow each Citrix user 30 Kbytes of bandwidth and there is adequate VPN capacity and licenses. Finally, ensure back-end Citrix server performance. All three can be the downfall to a very successful DR system.

On a more end-user focused note, ensure all core business applications are available and users understand which applications can be accessed remotely. For financial institutions, some limitations exist on what type of data can be accessed from none company premise – check with your compliance department.

It is days like today that Citrix proves its real value.