Other considerations

Speed is a hard thing to grasp. I mean, often times the "filling of a pipe" is done so using aggregate users of the SAN. So... perhaps a better way of looking at things is how many SANs using a singular client, can saturate the line. It should be fairly easy to saturate a line when considering requests over a multitude of clients.

It's possible that a SAN CAN saturate 16Gbit, for example, if using ONE client... but how many drives and what config was able to pull that off? Again, it sort of matters when considering how a particular SAN storage unit scales.

So things to consider:

1. Number of clients

2. Number of pathways

3. Number of drives

4. RAID level

And probably a lot more....

Microsoft abitrarily taxes VDI

One potential barrier is the cost of VDI. You see, Microsoft was forced to create an arbitrary tax on VDI through a new annual renewable license called the VDA. For every NON-Microsoft client that is doing remote desktop to a Windows host, you must purchase a VDA. The only way to avoid the tax is to use an all Microsoft approved infrastructure for VDI. I hear that EA users can exchange their unused CALs (which are "one time" fees) for an annual renewable VDA.

Basically what Microsoft has done is greatly raise the TCO of a typical VDI deployment. But of course, Microsoft will gladly "make a deal" with you with regard to the "tax"... so it may only be a tax against SMBs. We'll see.

The VDA is an anti-competitive practice btw. As it coerces companies to choose only Microsoft approved solutions.

Science anyone?

For, the problem is that science isn't being done. Science is all about experimentation and observation... but now, if somebody has a "title", hypothesis can be promoted to theory and theory can be promoted to fact. Personally, I think it is wise that scientists bring differences to the table... but let's also be willing to do the hard work again... hard... sometimes long and painstaking work... and instead of basing new work on bad "scientific" assumption made by those with "titles", let's at least be willing to call a hypothesis a hypothesis again...

Theories really need to have a lot of repeatable observable data ... and then... we need to remember, it's still only just a theory. When we build large ideological structures (thories and sadly, apparently, facts) on top of a foundation of mostly theory (or worse, hypotheses), then we make almost all observations null and void.

Science anyone?

Theories are NOT popular feelings or generally accepted ideas.... (repeat that 3 times everyone). Creationists are motivated by what they believe God has said. Non-creationists (often) are motivated to disprove God (sad, but very true). And sometimes, that leads to bad reasoning and jumping to conclusions without doing the hard work mentioned earlier.

For those peuedo-scientists that place much of their belief is something they mistakenly refer to as "Darwinism", I dare say (can't prove this) that if Darwin were alive today, he would chunk most if not all of what he wrote.... but in all fairness, most "Darwinists" don't even understand what Darwin wrote to begin with (they're kind of a cult IMHO).

IBM... tandem, etc...

IBM still holds Itanium as viable for DB2 deployment (potential migration for existing Oracle customers):

http://www-01.ibm.com/software/data/db2/lowerdatabasecosts/

Also, realize that to migrate all Tandem users (e.g. pretty much all of the banking industry) off to a different platform is difficult. Not saying it's impossible, but I would think that would mean the Itanium stays around for some time.

VDI is many things

VDI, is not remote desktop. It's really a Virtualized Desktop Infrastructure, that is, more like a "shell" that holds applications, the shell and the applications could be coming from almost anywhere. The idea is that the desktop is an abstraction of software that somehow can work together to provide an environment that feels the same as a dedicated OS desktop.

This works well for software that was designed to be multi-user and distributed. Unfortunately, most Windows applications were only designed to handle a single user, single platform mode. This is probably where most of the headaches come in .... and arguably is the biggest hindrance to VDI adoption.

The desktop I'm typing this from is NOT Windows based, nor are the applications. My email client runs on one host and by browser runs on a different host. Yet I can download a file via the browser and attach to email, likewise, I can click a web link in email and it will open up in my browser... this is more like what VDI is meant to be. Seamlessness between disparate applications running on different platforms (and it can change from platform to platform as needed).

So.. VDI is much more than centralized remote desktops... the scope is much much larger.

Biggest issue for me.. multimedia. For that, part of VDI, at least for now, requires a local piece running on the front end, which limits the flexibility of VDI and as mentioned, weakens its security as well. Network introduced too much variable latency preventing most common video formats from working well (you end up with audio/video sync issues, which are too painful for most to accept).

Fond memories

Owned the Rev A myself. And did have the external monitor as well as the 300 baud Osborne pulse dialing modem that fit into the floppy holder slot by the port. The Rev A had double density single sided drives where the original had single density. So everyone cut the notch in their floppies so we could turn them over and use the backside. My Osborne I ran: Turbo Pascal, COBOL, LISP, C, Z80 Macro Assembler in addition to the normal package and the plethora of add ons through the many BBS CP/M sites. At school, armed with a 300 baud modem, I did my mainframe work via Wordstar and uploaded it to save valuable "dollar" allotments on the school's mainframe. I was the envy of my dorm since 128 scrollable display is good enough to display the majority of mainframe output which was formatted for a maximum 132 character line printer. I made my own modifications to OSWYLBUR to handle the strange Osborne I modem... and many of us replace the CP/M shell with ZCPR, a command replacement with more features. I even hacked in a pulse dialing modem routine in place of the built in DIR command, since most people used a directory listing program from disk instead. I also programmed a game using the Software Toolworks C compiler where you flew around the screen and turned asterisks into boxes. The asterisks would kill you if you ran into them and the boxes were like walls, so as you played your ability to move about the screen decreased. What fun! I also wrote a mainframe 370 assembler in macro Z80 assembler. This allowed me to do a lot of my labs without using valuable compute time... just had to upload the final product. In high school, I developed a text adventure game (ala Infocom) where you had to solve chemistry problems to get through obstacles. In my junior year of college I wrote a small BBS in assembler for my Technical Writing class.

Great machine... I wish I had never given it away. It was very useful. It was fun keeping my dorm mates up all night as they listen to my TTX 1014 daisy wheel printer typing away....

SaaS or other services and software freedom

Having heard RMS many, many, many times... he'll tell you that the point of software freedom is to protect software where distributed. His example deals with a printer and the fact that the "intellectual property" software, because it was not available, rendered the printer useless after the vendor ceased support. Valid point.

But he'll tell you that you can use and modify GPL code and use it internally as long as it is not distributed... again, to keep the end user from ending up with a "useless" system someday.

So... now we have services. The problem is... what is delivered to the end user? Software? A device with software? It's harder to see....

But... let's say the service software was distributed and the the provider "dies". The customer now has the ability to setup their own "service" even if just for servicing their own. Otherwise, the service they had just became "useless"... and you end up with a very similar problem to the software freedom issue that RMS likes to espouse.

Something to consider...