Healthcare.gov: Another Example of IT Being Taken For Granted

Share on LinkedIn Share on Twitter Share on Facebook Share by Mail
Posted

Numerous other sites and media outlets can discuss the politics related to Healthcare.gov, but the design and build of the site has a more IT related project management issue. It seems that while the world relies on technology we still have yet to figure out that technology cannot be served half baked. To evoke a favorite phrase of mine: Proper Planning Prevents Piss Poor Performance.

Leadership

Let’s start with leadership. In order to have an effective project, the leadership needs to have a vision, have that vision be forward thinking/reaching, and strong in their managerial skills. The major break down in the leadership structure is the former CTO of the Department of Health And Human Services (HHS) and current US Chief Technology Officer/Assistant to the President, Todd Park. Mr. Park’s…let’s be honest…Todd’s leadership has been described as born out of Silicon Valley. The idea of one of the most powerful IT people in the world running their department fueled on no-sleep, pizza, and Red Bull (Mountain Dew for the purists) is ludicrous. These types of antics are allowed for leaders of companies like Facebook, Twitter, or Instagram, but not someone who’s developing one of the United States’ most complicated and scrutinized web based services. A month after go-live we still have major IT related issues with no clearly communicated path for repair. At this level there’s just no room for agile leadership and all nighters.

Contractors

The next target would be the major contractor, CGI Federal. Did no one do their homework on this company prior to hiring them? They have a litany of high dollar-government related site delays and failures. It’s the government’s prerogative to hire outside firms to build their sites and online services instead of using the large body of developers that already work for the fed. This decision is easily paralleled in today’s enterprise environment. Companies choose to go outside their walls to hire contractors to do work for them instead of relying on the full time people already in their employ. I cannot condemn this action as in some cases it makes sense. However, contractors need to be treated like a tool. A certain tool is chosen for a specific job, and if the tool has a history of not being able to complete the job properly a different tool should be chosen. The job cannot hinge on one tool. Perhaps the fed should have looked at CGI Federal’s Angie’s List reviews before hiring them.

Development

There are arguments in the developer world as to how to design a product: Agile vs. Water Fall. In some instances going 100% in either direction make sense. If you’re working on a low-level project or a spin-off idea and you’re merely looking to see where the rabbit hole goes, agile development can work for you. If you’re working with a large project with a laundry list of stakeholders and a specific set of requirements, it’s best to approach the design via Water Fall development. It seems that the HHS and the federal contractors had the worst mix of both. A government department trying to act as an agile development firm working with enterprise contracted development companies that work on deadlines and firm goals. A glaring example of the development process breakdown is found in the amount of time it took the HHS to get the site specifications to the various contractors. They could have mixed the development ideas by simply coming up with an initial list of “must haves” and then editing that list as the development process went on and it would have been a more successful venture. I’ll save a rant on the development processes theory for another entry, but suffice to say that this is a glaring example of both processes crumbling.

Expectation

The last issue to discuss is that of expectation. Thanks to the hard work of a number of developers over the past few years, technology has become easy and somewhat lazy. If you want to download a song or a movie and take it with you on the go it’s as simple as pulling up the media store on your mobile device which connects over a high bandwidth wireless connection and pay/download. The world has become so accustomed to instant service that we sometimes forget that the development of this type of transaction didn’t happen overnight. It really begins to show when people in positions of power can’t seem to grasp why the development of a large scale project can’t happen overnight. Everything else in their digital lives happens in an instant exactly when they want it to, so, why can’t the project they need done be accomplished in the same amount of time? Thanks to these types of expectations developers have been forced to turn in half completed work and claim the issues can be fixed in upgrades, service pack, patches, etc. When the product is the portal to something that’s going to be used by millions of people, we cannot allow the developer to follow in the footsteps of Apple and release a product before it meets the expectations of the users. This type of rushed work is what leads to disasters the likes NASA faced in Jan. of ’86.

Final Word

The problems that plagued Healthcare.gov are not limited to government IT related projects, but instead are a reflection of how the world continues to treat technology, its creation, and its support. Some may say that I’m over simplifying the ideas of getting the right people to lead the project, find the right people to build it, and make sure it works before you give it away but I say that those people are over-complicating the issue. I know some of you out there have dealt with these issues whether as a developer/designer or as a product purchaser. Feel free to vent below.

Citrix Tech Preview of OpenGL Support

Share on LinkedIn Share on Twitter Share on Facebook Share by Mail
Posted

 

 

OpenGL Logo  Citrix_logo

Last week Citrix quietly released a Tech Preview of an add-on to XenApp 6.5 which allows OpenGL support in the application as a service world. This is what was missing from our previous work, and now sends us back to the lab to test it out. Nothing to report on at the moment, but it looks like a friend of the show Thomas Poppelgaard has already had some hands on with the new software. Here’s hoping he demos some of it at the upcoming Nvidia GPU-TC.

UPDATE: I’ve gotten the plug-in installed on a demo system and have been very happy with initial results. I’m hoping we can get some more formal tests put together and record a vid of the results.

Citrix Blog Post: http://blogs.citrix.com/2013/02/22/how-many-users-can-share-a-gpu/#comment-73419

Poppelgaard’s Demo Vid: http://youtu.be/yH1vHAUSL98

Poppelgaard’s Write-up: http://www.poppelgaard.com/techpreview-citrix-xenapp-6-5-opengl-gpu-sharing

Nvidia’s GPU-TC: http://www.gputechconf.com/page/home.html

Tweak the graphics

Share on LinkedIn Share on Twitter Share on Facebook Share by Mail
Posted

During the latter part of the GPU pass-through project I began grasping at straws to try and tweak the system into better performance. In that search, I came across two items of note:

 

1. An old project from the late 90’s called GLDirect:

http://sourceforge.net/projects/gldirect/

The idea was simple (kinda): convert the OpenGL calls to DirectX calls. From my forum trolling I found that for certain games it seemed to work, and even sped up some peoples gaming rigs. I can say that for this project it didn’t workout. At least it didn’t in its current form. With some work, it may be something to aid us OpenGL users. I spoke to the creator of GLDirect program (for privacy reasons I’m omitting his name) and he would love to get back on the horse and update the software, but this isn’t some overnight re-code. This would be a major project, and would need funding. I wish I could assist but other than mentioning it on my site, I don’t have the cash to float him. Perhaps some multi-billion dollar software company could hire him to integrate his product into theirs….I’m looking at you Citrix.

 

2. HDX Progressive display

Apparently Citrix doesn’t want you to over exert your GPU so they throttle its output. There’s been a number of articles written about this so I’ll let them take it from here. My only comment is I tried this on a system with GPU pass through capabilities as well as a system with Win Server 2008 R2 installed on the bare metal and couldn’t see a difference in the graphical output for my apps. Let me know if you have something I could try to really see it in action.

http://blogs.citrix.com/2011/05/30/hdx-progressive-display-dont-forget-to-turn-it-on/

http://forums.citrix.com/thread.jspa?threadID=293310

 

Cheers.

Citrix GPU Pass-through and OpenGL

Share on LinkedIn Share on Twitter Share on Facebook Share by Mail
Posted

Outside of the GPU pass-through video project I’ve made, I began working with XenApp 6.5 back when it was still in beta. It’s been almost 2 years now, but I still remember the initial setup and testing of the HDX capabilities.

I’ll admit I should have taken more time to look behind the curtains and reviewed what resources were being used and how much, but I was so smitten with just the idea of moving my CAE (Computer Aided Engineering) apps to the cloud that I overlooked some very important information.

In the third video of the series I had to do some back end research on XenApps capabilities and the GPU driver calls of my software. As I found out AutoDesk is the only major CAE player that uses DirectX and everyone else (Solidworks, SIEMENS NX, Catia, Pro-E/Creo, ArcGIS) uses OpenGL for their graphics calls. I also realized that the OpenGl graphics calls were being handled by the CPU in the XenApp environment and not the nice fancy Nvidia GPUs.

I was able to run the software, but in a limited graphical capacity. The larger the file, more complicated the assemblies, or the higher the realism was turned on the more you could see the separation in the armor.

I went back to the various Citrix forums and did some digging. One thread:

http://forums.citrix.com/thread.jspa?threadID=295641

had a poster (Thomas Kotzing) that mentioned Citrix had some level of OpenGL support in the XenApp 6.5 beta but removed it and coded into XenDesktop. His web site says that he doesn’t work for Citrix so I cannot verify what he says. If it is true, I think it was a poor move on Citrix’s part.

In looking at the new products from Citrix (Project Avalon) and Nvidia (The GRID), I get the sensation that the Virtual Desktop is getting more attention than the Virtual Apps are nowadays. The systems admin inside of me (who eerily sounds like Gollum) keeps mumbling that from our perspective virtual desktops are the same as physical desktops. You still have to format the GUI to some company standard, and generate the “Golden Image” to be passed around. I already do that for my physical systems. @Citrix and @Nvidia why can’t I get the same level of service for my V-apps as you guys are making for V-desktops?

I mentioned this to a colleague at Nvidia, and while they couldn’t make any official statement, they did say that 2013 still has a lot to offer.

The Nvidia GPU Tech Conf. is in March so maybe something will come from that. We’ll see. For now, I still recommend grabbing some hardware if you got it, a demo license from Citrix, and testing out some of your own files to see what works and what crashes.

Cheers.

Nvidia Quadro K5000 does not support GPU pass-through

Share on LinkedIn Share on Twitter Share on Facebook Share by Mail
Posted

 

BREAKING NEWS

After a burst of e-mails with engineers from Nvidia, I’ve been informed that the flagship GPU of their new Kepler based cards, the Quadro K5000, is NOT supported for GPU pass-through. Without the GPU pass-through, the specific recipe of software I’m using for my cloud test isn’t going to be possible. Luckily after we were all brought up to speed, my contacts are providing a Fermi based Quadro 6000 to finish the test.

 

My current project is an attempt at generating data that would shed light on the number of users it takes to fatigue a XenApp graphic intensive server. Some of the hardware dedicated to this project were two Nvidia Quadro GPUs: the Fermi based 5000 and the Kepler based K5000.

Recently, hardware issues with the VM using the K5000 GPU tied to it led me to reach out to various support engineers at Citrix and Nvidia. The Windows Server 2008 R2 (x64) VM would not enable the K5000 GPU. The hardware manager would simply list “Error 43” and force the generic graphics to take over. This is not unusual behavior and at times simply powering down the host/physical server and then bringing it all back up enables the GPU properly in the VM. Unfortunately, when I attempted this the entire server would crash and reboot the host.

One of my contacts pointed me toward the XenServer Hardware Compatibility List (HCL) to show me that the K5000 was not listed. I knew that it wasn’t on the HCL but assumed that it was too new for the list or the site hosting the list was yet to be updated. As the flagship GPU in the new Kepler line of Nvidia Quadro GPUs one would expect the same if not better performance than the Fermi based cards.

I then received official word from Nvidia: “We (Nvidia) do not support GPU pass-through on the Quadro K5000 product.” I did some digging and couldn’t find any documentation that mentioned this. (If I missed it please let me know and it will be posted as an attachment.)

They went on to say that “We have a product line that is focused on virtualization, and that is the VGX product.”

The VGX system is a bag of Pros and Cons at this point:

Pro

  • Dedicated to virtualization.
  • Essentially 2 Quadro K5000s on a single card (VGX K2 edition)
Excellent! A product that focuses on the issue and packs quite a punch
Cons
  • Passively cooled. You can’t (shouldn’t) put this in a workstation, but in a server with internal fans.
  • Unpublished release date
  • Unpublished price point
So right now, if you’re upgrading or you’re looking to stand up a new Citrix farm you either wait for the VGX or stick with the Fermi based GPUs.

 

What does the tech community out there think of this separation of workstation and virtualization? Is this a good thing because we now have dedicated hardware etc. or is it a loss of ability to move to the cloud with pre-purchased hardware?

 

How many of you out there bought this card hoping to do virtualization just like this?

 

Let me know your thoughts.