Tuesday, April 14, 2009

This does completely change the game!

I have been meaning to explain and comment on the ever growing popularity of Virtualization or Virtual Machines (VMs) for the last two years. A recently announced innovation in cloud computing, secretly under development over the last seven years by Onlive.com, was unveiled and explained in detail during this years Game Developers Conference.

Such an innovative method of capitalizing on centralized processing and remote systems administrations has never been more effectively presented, nor more capable of taking advantage of the Internet's advancing average bandwidth, than Onlive's example application. Admittedly this video is describing an "On Demand Gaming" service; however, it is further illustrating a new form of high-end cloud computing, featuring real-time high definition graphic results. Furthermore, it is an example of a centralized computing environment which will dramatically capitalized on the substantial advantages of virtualized computing.

The innovation and results demonstrated in the presentation video below, are far more reaching than initially described. Onlive's dependency on virtualization and this groundbreaking, progressive, example of cloud computing, are more than just a glimpse of the future. The presented synergy of "cloud computing" and virtualization are a blatant leap forward and a genuine view of technology which must be considered and soon adopted by any progressive technology dependent institution or company.

It is a natural bridge to soon apply this technology towards instantly delivering and executing advanced, graphic demanding, state-of-the-art CAD (Computer aided design) applications for use within any architect, engineer or city planner's design and review process. Alternatively providing such professional applications could yield a system capable of unlimited, cloud based, clustered GPU results; simulating photo-realistic, full-motion computer generated output in real-time for any design project under development and review.

This is an opportunity that will LEVEL the playing field for even the smallest of A&E service firms which could benefit from high-end technology normally beyond their budgetary considerations or daily requirements. Drastically reducing typical cost associated with a local IT staffing and large departments. No software upgrades to worry about, license codes to manage, sophisticated hardware purchases to upgrade each year, or equipment leases to exchange, renew or turn in every two years.

The game has truly changed! Now you know why AutoDesk is one of Onlive's venture capitol partners to Onlive's unique process! Not to mention the inspiration AutoDesk will have to prepare their game authoring tools to take advantage of this new model in Game distribution.

Virtualization has been a fast expanding, emerging technology for consumers over the last couple of years; one which best exploits and better utilizes all available computing resources within any company, educational institution or ISP. It promises the ability to dramatically capitalize (or amortize) the huge repetitive and often underutilized investment in CPUs and related software and technology. When properly implemented Virtual Machines can offer HUGE cost savings for the end user. Helping to avoiding the surmounting annual cost associated with built in obsolescence and continued maintenance.

Virtualiztion drastically reduces any concerns or efforts spent upgrading and maintaining systems as well as their application and base OSs. Additionally, the technology leverages far fewer attached, physical servers in order to accomplish far more task. Furthermore these advantages and efficiencies may be applied on a company wide basis; not just in the "server rooms", additionally serving as a foundation for each end user's workstation throughout the company.

Despite all of these advantages, IT departments are, as usual, slow to adopt and implement this exciting progression in computing, cost savings and efficiency. The bottom line is, the cost of implementing such VM environment pays for itself in the first year of operation or less. Hesitations in utilizing VM's is more likely due to the lack of understanding its implementation and inability to fully calculate the financial advantageous which are crucial barriers for any institution or company to solve prior to risking innovative thinking.


Tony Richards said...

MMO Games stream at a rate of about 100kb/s and cost about $1.5 / user.

OnLive low resolution is about 10x this much.

MMO Games can host about 50 players per computer. OnLive? Only 5 players per computer (or only 1 player if they buy a cheaper computer).

Overall, the wholesale cost for such a service will run about 10x more than your typical MMO game.

World of Warcraft is making money hand-over-fist with their $14 / month subscription, so possibly they could reduce their price to $7 / month and still make a profit.

OnLive would need to be retail priced at $70 / month in order to make a profit... and that's before they have any premium content.

Remember, WoW is a game. OnLive is only a game delivery service, so you'll still have to figure out some way to provide content for the delivery system.

In the end, you'll be able to equate this as a $70 / month for basic service (Pong, Checkers, Hearts, or possibly some old classics like Tetris, etc) and then premium services where you get a choice of two or three games a month like NetFlix will cost you an additional $20 - $40 / month.

The only people willing to put out that kind of money to play games are hard-core gamers that spend an average of $100 / month on hardware and games and they prefer to own their own rigs.

I'm not saying it won't happen... people used to complain about paying money for cable when broadcast TV is free, but tons of people pay fairly high premiums for cable services these days.

But... I really hope it doesn't happen.

Do we want OnLive becoming the Microsoft of games, where you have to pay the "OnLive tax" if you want to publish / play games?

What next?

Just as the cable company charges you for their services, and still inundate you with 15 minutes of commercials for every 45 minutes of content, games will quickly follow suit.

No thanks.

This article covers a lot of this in more detail and also provides views from both sides (whereas my response has been fairly one sided).

M.Hulme said...

(Sorry for this comment deletion and republication! I tend to do that on each of my initial blog entires as well - my form of self-editing! Bad habit.)

Good points where related to cost, regarding many of the subscription models you mentioned. However, I do feel the advantages of Cloud Computing go a bit deeper than just these cost comparisons. But I'll touch on those less obvious impacts towards the end of this reply comment.

I was also hoping you might post the link you hyperlinked at the end of your comment to "this article". I couldn't access it, but do want to read anything you recommend. In all honesty I'm looking over many of these issues for a different industry all together, but the gaming examples serve well for my considerations. And I always respect your input.

First I'd like to bounce some thoughts back in forth with you pertaining to your subscription expense considerations; for instance, comparisons of a hosting scenario of (50) MMO client connection per server -vs- (5) individually virtualized gaming environments per server. Which does assume other rather preparations and large expenditures have already been made in advance by each end user.

Lets look further at comparisons of the estimated $100 per month as a base subscription, assuming the cloud computing monthly subscription provides access to desired premium content; while considering more tradition cost models, including upfront expenses for the gamer's hardware requirements. Lets also assume that we can only justify such a subscription because we are a customer that wants to play on the most recent generation of gaming consoles and PC gaming hardware titles.

Additionally, we'll assume in order to gain the "right" level of PC gaming hardware for this type of gamer, one would modestly spend between $1200 & $2200. This price does not including additional controls/peripherals or the monitor, but does provide for adequate removable media/ROMs, RAM, Processor, power supply, case, cooling, GPU, HDD, etc... In both cases, broadband internet will be needed (understanding the Cloud Computing solution does require a premium version to achieve 720p (1280x720) results.)

Therefore any enduser responsible for their own CPU solution has already contributed $100 per month towards either 11 or 22 of their usage, depending on the quality of system they choose to purchase. (And assuming a cloud user already has an existing older laptop or desktop capable of minimum cloud reqs. or $250 for a new Netbook.) In the case of the cloud opportunity, every users gains access to "high end" hardware from day one. Additionally, this cloud based hardware's "quality", will be expected to improve and update on a regular or game/application relevant basis. As apposed to whenever each enduser can afford or justify new upgrades 11 or 22 months form now. (Which would be a limited option in the case of typical laptops.)

And while the local CPU user's internet subscription may be cheaper if they chose a lower bandwidth, they must also purchase, retrieve and maintain all necessary software and local content prior to establishing each and every game play option. Obviously typical local CPU users will still be required to download, store and update all applications and databases to their local HDD; retrieved at required intervals either directly from the developer, or by purchasing typical pre-packaged, retail, boxed software; along with the purchase and maintenance of all future updates. Whereas this model has a new type of very advanced flexibility within a cloud based environment, administered and delivered in the way Onlive describes. (I'll expand on this centrally located DB more towards the end of my comment.)

When I constructed and utilized a custom, shared GPU system (based on a Stanford developed API called WireGL & later commercially available & supported at no cost as "Chromium") several years ago, I utilizing 6 individual, racked CPUs. (Full motion illustration @ http://www.michaelhulme.com/rackvideo24.html) The Chromium API allowed me to bucket the geometry being displayed and reassembled via the final CPU and load balance the resulting, complex 3D simulation at will, between the 4 slave CPUs required to crunch the GPU based task. Admittedly, an inverse opportunity for CPU load balancing is typical prescribed through current Virtualization management (1 to 6 -vs- 6 to 1).

However, in case of either a shared or individual, locally hosted, CPU computing environment, one fact is true; during an initial 22 month period of time; while the original investments in hardware is amortized, the capability of GPUs will further advances in overall performance by 400%. Given the history of GPU performance advances typically 2x outperforming Moores law for CPU transistor counts, etc. Also true, is the fact that the gamer invested in a local system, can simply upgrade to the next generation of GPU for only the isolated cost of the GPU hardware. I took advantage of this upgrading opportunity when I had the locally based shared computing solution, as my previous video link set out to illustrate to the board members I sought and obtained initial funding from. (http://www.michaelhulme.com/rackvideo24.html)

As we look at these two arrangements, based purely on hardware cost for the moment, we have to also ask ourselves how many gamers (or should I say, "typical American consumers", much less corporations, who typically depend on leases, deepritiations and operational write-offs.) save up $2200 cash for each CPU purchase prior to buying? Versus, requiring the use of credit? The point being, if you are the type of gamer or end user who wants the best hardware from the start; does this also mean you are the type who always has the capitol or cash to pay for it upfront? Probably no real useful correlation can be made here, and not one a game developer should have to be bothered with this issue, regardless. The bottom line is, not having to worry about such initial hardware cost isn't such a bad deal for the consumer or the developer.

Beyond the advantage of not requiring up-front capitol for high-end hardware; is the convenience and potential advancements when provided with the knowledge that the full power of an ever progressing, state-of-the-art gaming system, is made available to you from any location equipped with a 1mpbs or better connection. Impressive possibilities here... A typical Netbook just became a "Super Computer"!

And as apposed to spending an additional $380+ dollars every 22 months for an upgraded GPU (assuming you don't mind keeping the old CPU, OS, HDD & RAM) you could simply rest assured that any new gaming application hosted on the Cloud Environment will be accompanied with the required hardware necessary to delivery a 720p result at time of product release. Versus, whenever you decide or "save-up" for and initiate a system upgrade.

Now try to picture what a game developer, knowing a powerful, virtualized, cloud computing environment had been widely adopted, could initiate specifically for those users to take advantage of. And allowing me to digress back to technologies I mentioned earlier, such as the shared computing power of the WireGL / Chromium system. Exploitations could be designed for battles and higher level experiences and strategies within a cloud computed MMO sessions providing, for instances; cut scenes beyond those available to single, local, CPU users but directly based on the previous battles outcome and environment deformation which just occurred. Post battle recaps featuring Cinematic quality, overviews displaying environmental realism and effects on a level not locally possible, with aggregated damage, troop and vehicle counts along with close-up realistic appearances acted out by remaining cast members or within the altered levels and game venues. In addition, pre-battle intros could also be scripted in a similar way showing the over all battle environment, allowing for deeper strategy development via an at-a-glance, cinematic, level of details that might not be possible during game play without say a $4400 or even better a $440,0000 machine.

Consider download the "Kill Zones 2 - Behind the Bullet", interactive demonstration from the PS3's Sony Online Store, and select the fourth commentary from the "lead technical programmer" while you interact with the realtime content, for other inspirations in this areas of game development. (Regardless of this discussion, you really must see this 110 meg Sony PS3 interactive downloadable demo! It is exquisite!) Understanding this content advancement opportunity may not be immediately prudent for Indie development, as much as for those involved in more liable 3+ years 100+ person staffed, multimillion (100+) dollar budgeted productions, requiring insane returns and consumer appetites. But of course this multi-billion dollar gaming industry / machine is going to continue to look for differentiators such as this "over the top" level of sophisticated development and hardware exploitation. Nice name for a new gaming company, "Over The Top"!

Now picture one more leap, not as related to end-user cost so much as their game play experience! Developers soon learn to move well beyond the before mentioned cinematic level cut scene exploits and more advanced pre-play & post-play strategic game environment overviews, and soon realize via this cloud environment's API they have access to a "super-computing" environment. Immediately available to all of their Onlive based gamers! Now if this newly aware developer could just reasonably (without the burden of developing a new engine or other extreme R&D cost,) develop on top of an existing engine and database standard a platform ready to exploit such power. A platform of gaming which requires multi-terabyte sized environments; having no restrictions where related to the complexity of content benchmarked by a single DVD ROM or BluRay or installation media, or base installation download sizes; much less, local, consumer level GPU & CPU processing power. Everything exist on the cloud, and with a little creative utilization of the Onlive API, it all just so happens to exploit existing database standards which were previously authored for just such a massively shared, scaled and interactive database thanks to that of the U.S. tax payers and our DODs thirst for flight simulation and special automated force (SAF) training.

Enter solutions such as "CDB", an open standard (http://www.presagis.com/products/standards/details/cdb/) which provides real-time visualization directly from "source data", thereby eliminating an order of magnitude of the cost associated with labor typically spent messaging source data through intensive content creation teams and budgets, in order to gear the results for local 1-4 CPU visualization on a single users machine. However, advantages garnered from standards such as CDB assume you have the multiple CPUs, back-plain bandwidth and RAM neccesary to subdivide and load balance the run-time authoring and computations required. Fortunately, within a quantified, cloud based, shared computing, virtulized environment, subscribed to by any end users, this is now an option.
(CDB user group meeting (If you happen to be in Europe this spring!): http://www.presagis.com/about_us/events/listing/itec_2009_cdb_user_group/)

The bottom line is, I feel confident such advanced computing offerings being made available to the general consumer level, for only $100 (est) per month, will ultimately yield an entirely new business, development, purchasing, and computing model; which is very hard to even imagine, but I do enjoy trying!

Given how frustrating such cloud based, virtualized advancement might be to prepare for or combat on the indie game development front, perhaps such possibilities are not currently a relevant issue or consideration? In other words, it is just not a possibility that can yet be approached by the indie development community. Or is this the perfect opportunity to not only level the playing filed, but for indie to set the new standards in MMO?

Conversely, perhaps for the moment, such progressive possibilities may only overlap with those large 100+ man, formal game production efforts, taking place over the course of a couple of years and supported by 100+ million dollars of capitol. It is perhaps a very fractured market consideration and opportunity at the moment; but one that will none the less, ultimately, have a HUGE impact on all areas of code development, content creation and end user experience in this industry. The internet has assured us of this influence weather our consumers or developers are aware of how, or can prepare and be the first to innovate, this emergent technology's impending influence is no longer speculative.