Through the Interface

24 responses to “Moving to the Cloud”

1. 

   Adam Davis

   March 30, 2012 at 4:01 pm

   I too see it as inevitable.

   I think it depends on the service offered. The idea that I could do some computation intensive processes online and come back sometime later to view the results is compelling.

   However, working with a product constantly via the cloud is less compelling. We have to consider the fact that it’s not the headline SLA of Autodesk or some high end cloud supplier but is the lowest common denominator of ALL the SLA’s of ALL the infrastructure between you and the service. In an age of distributed working that means your flaky home connection that probably has an SLA that is meaningless.

   In a strange twist of unintended consequences it could mean we all have to work in offices again where the core infrastructure is sufficiently robust to utilise the service thereby negating some of the advantages of cloud services…

   I see Autodesk is already trying it out with online help only. That’s fine but to not install it locally or have it available on the DVD is a step too far. We have a lot of clients who use AutoCAD on ships – not much core infrastructure there.

   I was concerned that we drive too fast to cloud only services leaving truly (or intentionally) disconnected users high and dry. As the case above shows it’s happened already.

   Reply
2. 

   Massimo

   March 30, 2012 at 4:50 pm

   I second Adam about the risk of moving too fast to the cloud while infrastructures are still lagging behind.
   I live in a modern and civilized country (Italy), but it will take years before connection will be granted for everyone and many more before high-speed connection will be widely available to allow on-line computing to be used on a daily base.

   Nevertheless, most of the cloud advantages listed above, can be achieved locally by other means, like virtualization.
   In this respect, given my 30 years IT experience, I see Information Technology sometime repeat itself, every 10 or 15 years the paradigm shifts between centralization (mainframes, terminal services, virtualization) and delocalization (desktop pc, notebooks, smartphones).
   Different hardware, but same concepts aimed to different needs, I believe the error is to thing one paradigm is better than the other and cannot get along together.

   Reply
3. 

   Mike

   March 30, 2012 at 7:59 pm

   IMHO the "cloud" is just another tool that can be utilized in addition with what you are already using, and will never replace locally installed software / services. Having access to cloud services for intensive computation or for collaboration makes sense, but moving your entire work environment and data into the cloud does not.

   Yes servers can crash and cause you downtime and lost data, but if you are running things correctly there should be redundancy and backups that will shorten or even eliminate downtime depending on configuration. The best of both worlds is to utilize both local and cloud applications.

   If for whatever reason I do not have a connection to the internet and it means that all of my data and applications are unavailable, this is unaccepatable.

   All it will take is for a serious local connection outage or loss of data in the cloud due to a provider mistake (or hacker) for people to see that the cloud can not be the sole provider for services.

   Each computer / server in a massive office can fail independently, and often there are spares available to replace the failed unit. If someone outside accidentally backhoes through my fiber and kills the internet connection which means that now 1000 people can not do their work from this one point of failure, unacceptable.

   Reply
4. 

   Kean Walmsley

   March 31, 2012 at 9:23 pm

   Thanks for the comments, so far (keep them coming! :-).

   I suppose I consider myself an optimist (or maybe a futurist?) regarding the required infrastructure. Just as PCs weren't capable enough to run CAD when they first hit the market, the march of progress (in this case measured by Moore's Law) meant it eventually became feasible.

   In many ways I think we're in a similar situation, right now in that (for instance) internet access isn't as pervasive as it needs to (and will one day) be. In time we'll also have redundant means of access - if one goes down, another can kick in. Some people have that today with WiFi and 3/4G (although we may well end up with something quite different, technology-wise).

   I also think that we will eventually establish a more flexible approach to running software: in the cases where we really have a dumb terminal (low local compute power), it will all run in the cloud and do the equivalent of streaming pixels. In the case where there are more usable local compute resources (and a smart phone would even fit into that category) more processing could occur locally. It's going to be some time before we have code that can easily shift between cloud-based and local cores, depending on availability, but my bet is that it'll happen sometime in the next decade. We'll see! 🙂

   Kean

   Reply
5. 

   Theo

   April 3, 2012 at 2:50 pm

   Reliability is the keyword.
   i've been toying with AutoCAD 2013 for a few days now, but everytime i try to even just sign in to the 360 environment, AutoCAD crashes, rendering the service unavailable and hence, useless. Although i am a firm believer in the advantages of the cloud and the possibilities it may offer in the future that go well beyond the mere storage of data in order to make it accessible from everywhere you go, at this point i am not yet convinced that AutoCAD is ready & able for this. Moreover i would have liked for AutoCAD to put some more effort in the multithreading/multicore/multiCPU technology which i feel has been more or less forgotten or abandoned while chasing clouds....
   To many things can still go horribly wrong with today's internet connections as well, WiFi is often not offering nearly enough bandwidth and 4G is still a dream of the future. Over the past years I have seen and heard many promises of a Brand New Future, only to realize that they didn't really deliver once they had become reality.

   Reply
6. 

   Dave Ault

   April 3, 2012 at 5:11 pm

   From the very word go not one major cloud proponent has addressed just how they will make all the infrastructure they do not own or control guarantee service levels to subscribers. They do not clearly spell out all the costs both up front to the buyer and then in ancillary costs to make it all work. They do not state how your customers with whom you have signed confidentiality agreements are going to allow you to breach security by going on an insecure web to work. Data ownership. The list goes on and on.

   Once again we have here an example of how wonderfull it will all be with instructions to not look at the man behind the curtain as in Wizard of Oz fame. Wonder what the next big software buzzword gotta have thing will be when this cloud stuff fails in real life use for all but the simplest of cad endeavors.

   Would you care to address in clear and concise terms exactly what Autodesk is bringing to the table considering my first paragraph?

   Reply
7. 

   Kean Walmsley

   April 3, 2012 at 6:56 pm

   I think perhaps I need to clarify my role, here. I'm not Autodesk's cloud spokesperson. In this particular situation I'm helping educate our development community on industry trends. Software developers have to effectively anticipate (or, at least, harness) technological shifts in order to maintain successful businesses.

   As I've speculated (and while I might suggest it's reasonably well-informed, it does remain speculation) within the next decade (or two, at the outside) we will reach a point where most software is delivered via the cloud. Where it gets run will depend on local compute resources, but the existing model of local installation is - in the main-part - going away. That's not to say that I know of any plans for Autodesk to stop allowing software installation anytime soon - we'll probably provide it as long as a significant number of customers request it, I imagine - but that's the way things are generally going in the software industry.

   There are, of course, many questions that need to be addressed, in the meantime. Questions around security, privacy, service levels, etc. If they're not addressed to the satisfaction of our customers, they'll continue to work as they do today (or possibly invest in private cloud infrastructures, etc.).

   Anyway - I'm not trying to convince anyone of the correctness of this "vision", nor that we (Autodesk) are necessarily the best providers of cloud-based technology to our customers (although I of course hope this ends up being the case). This series of posts is intended to educate software developers on how to start moving software capabilities to the cloud, where it makes sense for them to do so: knowledge which I expect to be beneficial for many people out there.

   Please do keep the comments coming, everyone. I'm very happy to see peoples' opinions of these shifts relative to our industry. I won't agree with all of them, but then neither will I expect you to agree with mine. 🙂

   Kean

   Reply
8. 

   Dave Ault

   April 3, 2012 at 7:17 pm

   Thats a pretty common sense answer. I can see how quickly technology is changing and it affects everything around us. My biggest fear here besides the never ending holy grail of sufficient bandwidth to make this work is that the bad guys are figuring out how to hack almost as fast as the good guys prevent it. Dept of Defense and NSA for instance regard the public web as being incapable of security. I watch this IPad and IPhone stuff and how quickly bandwidth is sucked up. I don't see ISP's ever getting to the point where they would be reliable for serious CAD creation because they are going to fund infrastructure to cater to the lowest common denominator where the majority of their profits lie and not intensive users like CAD would be.

   You developers really have your work cut out for you to make the cloud for CAD happen beyond a superficial level.

   Reply
9. 

   James Maeding

   April 3, 2012 at 9:51 pm

   Kean, What I am not clear on is how Autodesk would handle customization, or let's say "add-ons".
   My company treats AutoCad/C3D as a graphical OS, and does all kinds of things that are not part of the OOTB system. We have several folders of tools, plot configs, and so on. I have worked on AutoCad since version 10, and now (CAD) manage over 60 users in a few offices. I've watched Autodesk change things about AutoCad each release, and my conclusion is they now have a mess on their hands. It seems they never sat down and decided where settings should go, but only added on as new features arose. We now have a mix of registry, xmls, daisy chained menus, and workspaces embedded in menus.
   I am still seeing confusion on Autodesk's part about basic menu configuration. They do not have a recommended basic setup that is consistent. The default profiles confuse the heck out of people if they compare them.
   So if we were to run acad via a service, how would Autodesk accomodate this? Must our tools be on the cloud? I consider this the most important issue to this cloud idea, similar to how important keeping Lisp is to Autocad. You handle it wrong, and every other cad product becomes extremely attractive. I think there needs to be a lot of discussion about what cad managers consider right and wrong handling. The problem is Autodesk has gotten most of these kinds of issues very wrong in recent years. It seems like they just do not have time to tryly investigate real usage at real companies. I would hear about it if they did, as cad managers in southern california stick together and talk. Autodesk has not shown the needed interest so far to define the real challenges IMO.

   Reply
10. 

    Kean Walmsley

    April 3, 2012 at 10:19 pm

    James,

    I'd rather this didn't become a discussion about how Autodesk might take specific products to the cloud (as much as anything I'm really not at liberty to discuss such plans publicly).

    Of course it's a "paradigm shift" (yes, I put that in quotes because I dislike the term) - and we inevitably have to balance legacy issues with the opportunities presented by this brave new world - and there's always risk involved.

    Ultimately we're committed to making our products strong platforms - both for independent software vendors and for internal customization & development - so we'll be working hard to make this work.

    Regards,

    Kean

    Reply
11. 

    James Maeding

    April 5, 2012 at 8:44 pm

    I understand, but until we hit details, we are really "thinking" about moving, not moving, wouldn't you say?

    Reply
12. 

    Kean Walmsley

    April 5, 2012 at 9:39 pm

    James,

    Here's what I'd say...

    This series of posts is to raise awareness of a fundamental shift that's happening in the software industry, and to help educate AutoCAD developers about how they can already start to harness the technology.

    It is not about specific Autodesk products moving to the cloud.

    I hope this clarifies,

    Kean

    Reply
13. 

    james maeding

    April 6, 2012 at 5:00 am

    Got it, guess you can see we are hungry for the details whenever they emerge. Thx foryour thoughts so far.

    Reply
14. 

    JGGerth

    April 6, 2012 at 6:27 pm

    Kean,

    Warning: wall of text ahead. I appreciate that you are not the 'cloud spokesperson' for Adesk, but I think that the points you made, while addressed to the developer community, need a response.

    While vendors may be excited about moving to the cloud and offering software as a service, the rationale is not in the least a benefit to the customer. The real reasons are to maximize revenue to the vendor irregardless of quality of the product, and to minimize the control customers have over their data- transferring that control to the vendor and encouraging customer lock-in. Especially given Autodesk's published ToS for 360 - there is zero reason to entrust any customer design information to their cloud services.

    As to the specific points you mention, the politest thing to say is 'that happens not to be the case'.

    Performance: you specifically mention chunking up data and paralleization as benefits to the cloud. It's important to point out that this is not a feature specific to 'teh cloud', and could be better addressed using private grid computing within an organization. There's no reason (from a customer's perspective) that the rendering actions should happen elsewhere, when there are spare CPU cycles within the company, and inside the companies firewall. Having built more than one render farm, and a continuous user of BOINC for the past 4 years, I know that such an approach is technically feasible, and far more secure and reliable than a remote hosted service.

    Scalability: I'll give you half a point on this, but it only applies to a limited subset of use cases, and is totally irrelevant to CAD and design usage. I may be able to provision extra computational resources, but staff, especially knowledgeable staff familiar with code and design requirements, are a far different story.

    Reliability: Sorry, but that's absolute horse hockey. If the resources I need are at the far end of a tenuous connection, filled with a multitude of point failures, reliability is non-existent. What's the merit of a remote service with 5-9s up-time when a backhoe tears through a line down the road? If I'm running desktop applications installed locally, the impact is minimal. SaS -- I'm out of business.

    Low Cost: That is almost assuredly incorrect. Given the realities of software vendors, and a multi-decade long effort to lock customers in, we as customers can expect that the total cost will only increase, and that the ability to move our proprietary design information outside of any given vendor's cloud will be severely limited. Heck, try moving a Civil3d Alignment to a competitive platform now, and see how much of the design intent goes with it. We as customers can only expect ongoing incremental charges, small individually but certainly larger than alternatives. Realistically, the best way for a customer to ensure his lowest total cost is to work with competitors, encourage competition between vendors, and hold vendors accountable for quality and performance of their software. Nothing in that is material to cloud computing or SaS.

    Following up with your 'Whens' - which look a lot more like "Whys"..

    Computation intensive: grid computing works, and is far more secure than a remote, untrusteable vendor's cloud services. Work with BOINC, and pass those specific tasks, whether fluid dynamics or rendering, off to the secretary's desktop, or the machine of the person tied up in meetings. Better, faster, more secure, and more reliable.

    Collaboration: Given the terms of service of Autodesk's offering, any CIO who does not block that service at the firewall should be terminated for incompetence, and any employee who utilizes has probably committed a firing offense under the corporate IT policies. WS may work, but is unusable. As is, there are better options, with more flexibility, reliability, and security. A self-hosted eRoom for example.

    Frequent changes: Certainly not a customer benefit - more of an excuse for a vendor to release alpha grade software as release and charge accordingly. Breaking customer applications, rendering their data unusable, skipping real testing in favor of customer testing -- not good for the people trying to get work done.

    Large data sets: OK, I’ll give you that one, sorta. but you're missing, or ignoring, the point that transfer of large data sets across an internet takes an inordinate amount of time. If have have a couple of terabytes of data to move coast-to-coast, FedEx is faster, and won't degrade everyone else in the company's ability to use the internet. Again, grid computing, using an internal BOINC setup, is a better solution for customers and end users.

    Pricing information -- maybe yes, that's out of my areas of concern. but immaterial to CAD and design.

    Customer intimacy: well, the big brother side is more important -- what I do with the product I’ve paid for is none or your business, and attempts to mine my interactions with the software are equivalent to spyware. There's an ethical leap to make, moving from the light side to the dark side, for any vendor to implement that.

    Technology abstraction: I'll give you that one too -- but it's pretty irrelevant to customers or users. Any benefit to the developer, is outweighed by the negatives to SaS for the customer.

    Device support: Trendy, but.... monoculture is bad, diversity is good, competition is good. Pretending that it, whatever 'it' is, will just work on any browser means that the specific abilities/features of my equipment are being ignored, and the software is working to the lowest common denominator.

    Reply
15. 

    Kean Walmsley

    April 6, 2012 at 7:37 pm

    JG,

    Thanks for taking the time to put this comment together. I'll try to keep my response shorter (it's Good Friday and we're lucky that it's a public holiday here in Switzerland).

    I'll start by saying that I respectfully disagree.

    Firstly, if the benefits are only for software vendors, why on earth do 100% of Fortune 500 CIOs believe the cloud is critical to their future IT infrastructure? (I made that statistic up, by the way, but I wouldn't be at all surprised if it turned out to be true. 🙂

    Beyond that, I'm really not going to get into a discussion on the specific points: it's clear you're coming at this from your own perspective. I believe this industry shift is happening and is going to continue (whether I like it or not). You either don't believe it's going to happen, or would rather it didn't. There are details related to the design software industry, etc., but frankly in the long run I don't see our issues as being so very different that we would do anything but lag the rest of the software by a matter of years (that's what happened with the shift to the PC, anyway).

    It's great you've taken the time to express your views, irrespective of whether they agree with mine. I might just set myself a reminder to check back on this blog post in 10 years time, to see which one of us turned out to be right. Assuming the internet and the planet still exist at that point, of course.

    Kean

    Reply
16. 

    JGGerth

    April 7, 2012 at 4:34 pm

    Sure, check back in 2022, and we may have a better picture. although.... IIRC SaS originally was offered, and failed, last century, so may take longer than a decade. And a lot of other IT fads and hype will have spun their way through to oblivion in that time frame. Few will have had any significant impact on the design industry, but a lot of IT and marketing hours will get spent on them.

    The products I'm designing today on the other hand, will still be functional and in use long after you've retired. barring planetary cataclysm...

    Reply
17. 

    James Maeding

    April 9, 2012 at 10:19 pm

    Kean,
    You can see that the three areas of what we call the cloud, must be clearly referred to when discussing the issues.
    There is virtualized desktops, data storage location, and software "running location"
    JG's comments picked on particular areas, so was specific.
    When you replied that you disagreed, you stayed general.
    You are saying "It" is going to happen, and each person can believe it will or not.
    But there is no one "It". The move to virtualized desktops is a perfect example. What did the CIO's have in mind when they said the move to the cloud was critical?
    I think we are actually all in agreement on many levels, then would disagree on how optimistic we are that certain critical details get worked out.
    You have to admit, if Autodesk relied on say, visual studio, for all coding, and MS said they were going to release a new version every year, and provide as SAS, you would not play that in a liberal manner. If it failed, and you were one in charge, what excuse would you have? So you will see that same conservitive nature on the side of people in charge of production.
    There is no reason to expect that to change, and it should not change, so I think those CIO's better check with reality before they get all optimistic on SAS.

    Reply
18. 

    Kean Walmsley

    April 9, 2012 at 10:36 pm

    James,

    I quite deliberately stayed general in my disagreement.

    I would love to be able to give specific predictions on what is going to happen and when, but I've found myself not to possess particularly strong powers of prediction. I completely missed predicting the turnaround in Apple's fortunes over the last decade, for instance, even in the desktop computing space.

    So I will repeat that I believe "it" will happen, and that "it" could come in a number of different forms. Time will tell exactly how and when.

    Also, please be careful not to translate what I'm saying in this blog into Autodesk product strategy. This message is very much developer-focused: I am *not* talking to product users with this series of posts.

    Readers (who actively do relevant development work and are considering where best to invest their time) can either choose to heed my advice and start thinking about the potential for building cloud-based services, or they can file it away for future consideration (or just plain ignore me) and see what happens.

    Kean

    Reply
19. 

    PLawton

    April 12, 2012 at 1:04 am

    Mr. W.:

    The cloud can solve a couple of problems that customers have, but many more problems that Autodesk has, prime among which is software piracy. As engineers, we NEVER need to render anything, and with Revit, 'synch to central' on a model hosted on Revit Server 1500 miles away can take 45 minutes. So running software as a remote service amounts to a bad business decision for our company, hands down.

    And I laughed out loud when you mentioned

    "Frequent change
    If you have applications that go through rapid release cycles, then update deployment/patching is likely to be a challenge for you."

    The 'rapid release cycle' is something driven purely by Autodesk (most users and IT folks have hated it for a decade), to maintain revenue flow thru the subscription program, so that reason is completely invalid.

    The cloud is being pushed as hard as it is primarily because it benefits Autodesk and its shareholders, and only secondarily for any potential benefits for the paying customer who uses the software.

    I also laughed when you mentioned that "you clearly have to trust your technology provider". This much we have learned: "Trust Autodesk to do what's best for Autodesk and for its shareholders."

    Reply
20. 

    Kean Walmsley

    April 12, 2012 at 1:32 am

    Mr L.:

    I'll start by repeating that this series of posts is not about Autodesk software moving to the cloud: it's intended for application developers who have code integrated with our various platforms.

    Certainly, if you're talking about moving .rvt files around, as you might today, then the cloud doesn't seem to make much sense. It might make more sense if multiple engineers were working on a genuinely centralised building information model, but that's besides the point (as this series is not about Autodesk software, after all).

    The benefits described are those we've communicated to - and have in my opinion been well received by - our external development community. For instance, there are several developers with frequently changing code and data (such a pricing data, that I've already listed as an example) who already benefit greatly from this style of system architecture.

    And my point about trust was quite deliberately made (although none of my points were intended to make you laugh, on balance). If you don't see the benefit of working in this way, then I don't expect you to be forced to.

    Regards,

    Kean

    Reply
21. 

    B2D2

    June 8, 2012 at 1:53 am

    Loved reading this post. Like to thank Kean and all those who responded. Cheers!

    Reply
22. 

    dave ea

    August 6, 2012 at 10:09 pm

    hi kean (and all those taking time to post their thoughts on this issue). great brain snacking here-- thank you. for me, the issue essentially boils down to the simple analogy of whether or not a company wants to own and drive their own cars or rely on a taxi service for all their driving needs (or some combination). the best decision seems to depend on each company's specific needs -- therefore i would never suggest that relying on a taxi service is always the better choice or suggest that the option for everyone to own their own cars might be going away. "cloud services" (offsite centralized computing managed by others) has actually been around a long time (no, we didn't call "the cloud" back in the 60's, but that's essentially what it was). maintaining absolute confidentiality/security of our data alone (bogus service provider assurances aside) makes the cloud a moot option for us right now, but others may not care who has access to their data (or where it goes) -- cloud services may be a very viable option for others where security/confidentiality is not an issue. my point though: if education is our core goal here in this thread, then it seems the responsible thing to do to present both sides of the issue: the benefits and profitabilities of this technology as well as its (very empirical) drawbacks and risks, so that an honest and competent assessment can be made for any organization's needs as to whether or not cloud technology is a helpful solution. you've given a very eloquent list of benefits and profitabilities, kean (thank you!). i for one would be very curious to read your thoughts on the drawbacks and risks.

    Reply
23. 

    Kean Walmsley

    August 6, 2012 at 10:58 pm

    Hi dave,

    Great to hear from you! 🙂

    This post (and its accompanying series) isn't really intended to encourage Autodesk customers to consider moving data to the cloud: it's encouraging software developers to consider opportunities for moving (parts of) their applications to the cloud. With the assumption that at some point in the next few decades they are going to see benefits from having made this investment.

    I completely understand that many Autodesk customers have various quite valid concerns about the use of the cloud with respect to their business. But developers need to anticipate industry trends and be ready for the point at which the customer benefits become compelling.

    Ultimately this is a pitch for (and a bet on) a certain future state. I'm not in a position to recommend whether an individual Autodesk customer is best served by the cloud or a local computing infrastructure, but I can say that developers who chose to ignore this trend are more likely to get a nasty surprise, sooner or later.

    Looking forward to chewing the fat on this topic at AU,

    Kean

    Reply
24. 

    dave ea

    August 8, 2012 at 9:37 pm

    hi Kean :o) thanks for taking time on this post (and this blog in general). it always keeps me thinking ahead and that in and of itself is a good thing. i think we're on the same page with regard to the benefit of any developer examining the prospect of porting his/her/their services and products to a cloud environment. i would only add that i think ti is equally important for any developer to consider the dangers and risks they face by doing so. looking forward to chatting at AU and thanks again for all your time on this blog. really good stuff here.

    Reply