Why Windows Azure Is Not Worth Investing In (For Developers)
It’s pretty obvious that I’m a Microsoft fanboy. This is the story of how they let me down today.
Why I Wanted To Learn Windows Azure
This afternoon I sat down to write a simple web service that would expose my cloud based Exchange calendar as public free/busy feed (webcal://tath.am/freebusy).
The hosting requirements:
- around 1 request every 5 minutes (0.003 requests/second)
- around 10kb of transfer per request (100 MB/month)
- no writeable storage or databases (it’s totally stateless)
Cloud hosting sounded like the perfect solution for this. I don’t care where it runs. I don’t care who or what else is running on the box. I don’t have any funky dependencies.
Whenever I go to write something for myself I use it as an opportunity to learn something new.
Today, it seemed like I was going to learn Windows Azure.
Why Azure Needs To Be Free For This Scenario
Being able to scale down a cloud hosting solution is just important as being able to scale it up. In fact, I’d go as far as suggesting that I think it’s even more important.
Right now, I can’t market myself as an Azure professional. I can’t get up at a community event and evangelise the product. I can’t answer questions about the product on mailing lists or forums. Giving me 25 hours of compute time to try the service isn’t going to let me learn the service.
The free availability of competing services is what lets somebody spin up a political statement in an hour. As well as getting a quick, free and suitably tiny hosting solution, Lachlan and Andy were sharpening their skills and building their own confidence with Heroku.
It’s not all about bringing users to Azure though – it also works in reverse. In the interests of being up to date with how ‘the other half’ live, I like to have a bit of a handle on what the Ruby world is up to. Recently I built http://tath.am/where so that I could publish a visualization of places that I’ve recently been and commonly go. This is a simple enough app that I figured it would be a good place to start some Ruby with however I initially dismissed this idea because I didn’t want to have to work out how to host it. In the end, the availability of Heroku actually brought me to the Ruby platform. Windows Azure is our opportunity to bring new devs to the .NET platform.
Attempt #1: Free via the Windows Azure Platform Introductory Special
My first point of call was http://www.microsoft.com/windowsazure/ where I spotted the introductory offer:
Sounds perfect! It’s free, and it has a teeny tiny bit of everything to let me play with it all.
Alas, this wasn’t the case:
That means I can only run my service for 48 minutes and 38 seconds per day.
Attempt #2: Free via MSDN benefits
As an MSDN subscriber, I have access to yet another “Introductory” offer. This offer includes 750 hours per month (basically the whole month, as 31 days = 744 hours) as well as way more storage and bandwidth than I would possibly use with this service.
The problem with this approach is that it only lasts for 8 months (assuming I disable auto-renew otherwise they just start charging me). After this time, my MSDN Ultimate subscription will only entitle me to 250 compute hours per month.
That means I can only run my service for 8 hours and 3 minutes per day.
Attempt #3: Just paying for it
Apparently my part of the internet is completely different to your part of the internet:
Once again, Microsoft have completely stuffed up their product globalization strategy.
Attempt #4: Pretending I’m American
Luckily, the country checks are easily bypassed thanks to no address verification service being implemented:
Before I hand over my credit card for “usage based billing”, let’s just estimate what this is going to cost me.
US$84.82 per month … for a web service.
The Solution
Once again, I’m back at Heroku. My web service will run with a single dyno (their compute unit) and happily tick away 24/7 for free. I already have one app on their infrastructure (http://tath.am/where) and now it sounds like I’ll be building my second.
Being realistic, I don’t expect many .NET devs to spend a weekend learning Ruby + Sinatra + Git + Heroku just to deploy a simple web service. Instead, I think they just won’t bother with Azure and will opt for traditional shared hosting or just not building the web service at all. The former is disappointing, the latter is terrifying.
Why Does This Make It Not Worth Investing In?
With the current pricing model, Windows Azure is destined for the same way that MapPoint went – a quiet solution in the corner that’s only used by larger corporates and lacks a viable ecosystem.
If you’re in the audience for this blog – Microsoft have made it clear that Windows Azure is not for you.
Unless you’re a Best Buy or Netflix contender and are looking to build out a prototype for the next PDC keynote, spend your time focussing on other solutions for now.
I’m Not The Only One
I agree completely. The MS view would be that hobbyists can play around on their local dev fabric/storage. Not much fun.
– Brad Curtis
Yeah I had high hopes for Azure initially, but now I’m not so sure. As you’ve pointed out, it doesn’t really scale down.
– Thomas Johansen
Azure could have been great with a trimmed down entry level package that was free. Sucks :-/
– Thomas Johansen
Agree, seems to be missing one of the big benefits of cloud virtualization – scale down as well as up.
– Jon Galloway
Yeah, it’s not very interesting at that price point. At least not to me.
– Dave Ward
dude, that totally sucks! Heroku is pretty sweet, though. If they’ve only made Azure as simple as that.
– Javier Lozano
well put, sad that azure’s out of reach for mere mortals
– Brendan Forster
Really seems they’re missing the point of hobby guys developing the Azure skills then taking it back to work with them.
– Andrew Tobin
Great post, keep it up I say. I have turned to GAE for similar projects for the same reasons. Good for everyone to speak up.
– Tarn Barford
Tatham starting trouble that’s for azure..
I lived on 1 Fuck You Way, Redmond. It was a great place to start off in, but I often found the neighbors with pickup truck car bodies were all to often bring down the neighborhood.
I think from memory that the AZURE service was deliberately priced out of the breadth crowd (Web devs) so that they could focus on Enterprise grade clients first. Stabilize that and then open it up to the unwashed masses..
That was what I was told once in a briefing. How true that is today is unclear but it seems to hint in that direction still.
Scott Barnes
April 5, 2010 at 12:42
Great feedback. I forwarded this to someone I know on the azure team.
Haacked
April 5, 2010 at 13:08
I would like to see a re-seller option, along the lines of what Javier Lozano added to the discussion on twitter ( http://twitter.com/jglozano/statuses/11620626333 ). This way Microsoft would only have to deal with “enterprise grade clients” for the time being.
Allowing someone to purchase a large quantity of Azure compute/storage/transaction time and re-sell it for an affordable price for low capacity users.
Nick Josevski
April 5, 2010 at 13:13
tatham,
Based on your description, I agree that it is not a good fit for your scenario, but ‘not worth investing in’ ?? common… feels like a bit of a sensationalistic title to me
based on that then, i would suspect your would claim that Amazon EC2 would not be worth investing in either then…
windows azure has been positioned since day 1 in the ‘virtual dedicated’ space not in the ‘shared hosting’ space. Many posts and threads have discussed this.
will there be a ‘shared hosting’ (fraction of 1 core / per user) offering in the future? maybe. Could someone built it on top of azure and resell fraction of core? maybe. But that’s not the point. Right now it is not and there are thousands of scenarios that can happily live with a minimum of 1 whole cpu.
As per learning Azure APIs, you have a full featured local dev environment; if you want to practice deployment etc. you have/will have 25h/month to play it; if you want to pay for it (and use a australian credit card), then wait another few days, azure is launching in April in Australia as announced several weeks ago.
you know my email address, feel free to contact me if you need more info or want to provide further feedback. But I clearly disagree with your blank statement ‘not worth investing in’
cheers mate
–gp
gp
April 5, 2010 at 18:58
Hey GP!
“I agree that it is not a good fit for your scenario, but ‘not worth investing in’ ??”
As a company who have always built themselves around the developer community, Windows Azure represents a reasonable step in the opposite direction. Right now I see Azure as being destined for the same way as MapPoint – a quiet solution in the corner that’s out of reach for a huge chunk of the market and doesn’t have a worthwhile ecosystem around it.
“will there be a ’shared hosting’ (fraction of 1 core / per user) offering in the future? maybe.”
I believe scale down is just as critical as scale up for a cloud hosting platform. How this is solved mostly irrelevant but I think it needs to happen, not just be a “maybe”.
“Right now it is not and there are thousands of scenarios that can happily live with a minimum of 1 whole cpu.”
Then there must be hundreds of thousands of scenarios that could happily live with a fraction of a CPU.
“As per learning Azure APIs, you have a full featured local dev environment”
I never used the term API; I’m talking about learning the platform. Writing demo ware mockups on my local environment isn’t going to take me anywhere useful beyond “hey look, it says Hello World”. Nevertheless, such a response was anticipated from Microsoft and already commented on by Brad Curtis: http://twitter.com/bradcurtis/statuses/11620554712
“you have/will have 25h/month to play it;”
Wow. 48 minutes per day! I’ll get to ration it like water.
“wait another few days, azure is launching in April in Australia as announced several weeks ago.”
How on earth the Azure team decided that their part of the internet has to be completely and utterly different to my part of the internet just leaves me in awe of their ability to completely screw their globalization strategy on practically every product they release. Whilst I can see the benefits of locally homed infrastructure certified under local schemes, these can be opt in scenarios. Many scenarios simply don’t care. On a tangent, the process of closing the CTP accounts weeks before launching the actual product leaves me even more dumbfounded.
– Tatham
Tatham Oddie
April 6, 2010 at 08:43
Nice post, I was going to do something similar. Hosting ASP.NET MVC really is horrible unless you spend the money for a dedicated box. I gave up doing my site in ASP.NET MVC and moved it to Ruby/Heroku just because it’s so easy to use (this site: http://monotouchexamples.com). I’d love to see something as simple as Heroku for ASP.NET but I doubt it will ever happen…
ChrisNTR
April 5, 2010 at 19:41
Tatham,
Fair comments in the context of what you’re trying to do, however, I think Microsoft’s strategy for Azure is not to make it a ‘bottom up’ adoption process in corporates, but rather ‘top down’. C-level people in an organisation will not decide to adopt cloud computing because their development teams are skilled up or keen on it – it’s a very big shift for those businesses, so the decision will most likely be driven by the senior decision makers, the business benefit and not purely the IT benefit.
Additionally, while MVPs such as yourself are there to help evangalise (and criticise
) a platform or technology you can only do that effectively once that platform or technology is mature enough. Right now I don’t see Azure as being at that stage in its lifecycle (they are still decomissioning CTP accounts – mine included, and they don’t have ‘nodes’ in each geographic region yet) so I’d expect that Microsoft will carry the load for evangalising it in the mean time – more so in regions such as Australia where a local cloud is something you’re more likely to see in the sky than on an IP address.
I think Microsoft is currently missing a big opportunity in being open to incubating online businesses on Azure because the more successful ones (Twitter anyone?) would give back far more over time than they would take during startup (and most likely cover the ones that don’t succeed). Maybe in the longer term Azure will extend out to this ‘hobbyist’ community – it would certainly benefit the platform – but I don’t see it as a core part of the current Azure strategy.
Cheers,
Simon.
P.S. – It’s Easter weekend – get out more!
Simon Waight
April 5, 2010 at 20:34
“so the decision will most likely be driven by the senior decision makers”
My audience for both this blog and my community engagements at large is largely developer focussed. My message is that these individuals and teams should not bother investing themselves. If the managerial team decided to move something to Azure, then that’s a business decision that IT can comply with just like them being asked to hack a barely supported Java search platform on top of a SQL + ASP.NET solution … ‘because it’s cheaper’.
– Tatham
Tatham Oddie
April 6, 2010 at 08:46
Tatham,
I’m glad/sorry to see that I hadn’t missed something. I ran into the same thing with Azure. It looks cool; MS is really pushing to us, and yet the price point is that it’s not possible for me to truly play with it (without spending way more money than I would want to.. just to learn.. who is this? Adobe?).
I think the fact they are pushing it to us Devs is something that Simon missed.. they are telling us it’s exciting and that we should consider using it and then the price point is way too high for me to even consider playing with it..
(Actually I’m changing jobs and the new company uses Azure, so I was hoping to get a head start on it before I start my new job, but I can’t at that price point).
Jay
Jay Kimble
April 6, 2010 at 00:25
“I’m changing jobs and the new company uses Azure, so I was hoping to get a head start on it before I start my new job, but I can’t at that price point”
Right!
“I think the fact they are pushing it to us Devs is something that Simon missed.. they are telling us it’s exciting and that we should consider using it and then the price point is way too high for me to even consider playing with it..”
Totally agree. If they want to pitch it to the dev community, make it accessible. If they want to pitch it business decision makers, get it out of our PDC keynotes and stop doing pointless developer roadshows about it.
– Tatham
Tatham Oddie
April 6, 2010 at 08:48
Tatham,
For the last year, I’ve been agitating for an Azzure billing threshold for developers, similar to that offered by Google’s app engine. See http://oakleafblog.blogspot.com/2009/09/lobbying-microsoft-for-azure-compute.html and http://oakleafblog.blogspot.com/2009/07/comparison-of-azure-and-google-app.html as examples. So far, I’ve seen no interest by Microsoft in adopting such a plan.
–rj
Roger Jennings
April 6, 2010 at 01:46
Hi Roger,
I think it’s interesting to note that Google App Engine started as a free offering and then added paid extensions on top where as Microsoft are working in reverse – try and charge everybody heavily up front and drop the prices for the segments that don’t adopt it. One approach seems calculated and the other seems like an experiment in measuring demand by fiddling with supply.
– Tatham
Tatham Oddie
April 6, 2010 at 09:10
Thanks for capturing your thoughts (and other’s) on this topic Tatham. It’s a good read and super valuable perspective for the Windows Azure Team here in Redmond – - my office must be just down the road from your street address
. I have shared it around as you would expect as it’s good for people to digest.
No doubt you and others would like to use Windows Azure for scenarios that require less than a full core (and pricing which reflects that), I am hearing that loud and clear. I don’t know who “gp” is above, but they do make some excellent points. Windows Azure “not worth investing in” is way too strong for a blanket statement – - for your particular use case today, you arguments are absolutely solid. For many other scenarios, our customers are telling us it’s well worth investing in.
Will Windows Azure evolve for you and others down the road and will we be able to more easily bring new devs to the .NET platform using Windows Azure? To that I’d stay tuned, you can check out my blog where I’ll provide updates to our roadmap and I’m always looking for perspectives on how we can do the right thing for our customers.
If there’s more you want to tell me or the team, don’t hesitate to shoot me mail and/or to share your feedback here, http://www.mygreatwindowsazureidea.com
Thanks again Tatham and others for your comments here, really apreciate it.
- Mike
Mike Wickstrand
Senior Director, Windows Azure Planning
Microsoft
http://blogs.msdn.com/wickstrand/
http://www.mygreatwindowsazureidea.com
Mike Wickstrand
April 6, 2010 at 05:55
Hi Mike,
Thanks for commenting here!
“Windows Azure “not worth investing in” is way too strong for a blanket statement”
Across the entire market, definitely. As I mentioned to Simon above, my audience for both this blog and my community engagements at large is heavily developer focussed. ‘Why Windows Azure Is Not Worth Investing In For Small Development Teams’ got pretty long so I trimmed it a bit.
(Un)fortunately this post is spreading a bit wider than that.
I look forward to seeing your developments in the space, and sincerely hope that Windows Azure doesn’t become the next MapPoint (great for PDC demos and useful to a select few companies but otherwise inaccessible and lacking community). The UserVoice site gives me increased confidence, particularly considering the content of the 1st, 2nd and 6th highest rating issues.
As an interim step, I think multiple roles per VM would significantly widen the market. I already pay $120 a month for my own managed server, and I could happy transfer that expense to a Windows Azure VM if it let me host multiple applications. Support for multiple roles per VM is critical to this though – hacking around with ASP.NET routing and one big über applicaton isn’t really an acceptable workaround.
Happy building!
– Tatham
Tatham Oddie
April 6, 2010 at 09:32
The reality is Cloud elastic computing does not actually exist, Aszure is no worse than others..
But to be fair is is realy just typical X.0 stuff, these are not mature services..
Big issues that need addressing before commercially viable:
a) inability to suspend services without charges.
b) compute needs to be identical to storeage, one does not need to buy discs, just storage, need to buy consumed processing not “servers”
c) stageing servies should be just that staging, not charegd as operational.
d) SLA’s need to be reworked
e) The concept of paying for a per instance of a free express database?
f) Need to haev a cron scheduler that maps to a),b)
dude
September 14, 2010 at 07:48
Already stated but remember that is what the local fabric is for, to mimic what you will get in the azure cloud.
Do you guys complain about having to develop against localhost as well or do you require a production web instance to build agaist?
Jason H
April 6, 2010 at 06:33
Hi Jason,
The local fabric is a perfectly valid /development/ model but it’s not going to take me past the “hey look, it says Hello World” scenario. Whilst I might personally be happy to invest time locally learning the product, it’s not going to capture the passing developer who just wants to whip up something small but still valuable.
“Do you guys complain about having to develop against localhost as well or do you require a production web instance to build agaist?”
No – localhost is fine for development, but after that I have to find somewhere to actually deploy it to.
– Tatham
Tatham Oddie
April 6, 2010 at 09:01
I have been posting similar thoughts a while ago about “tactical scaled down apps” at http://vermorel.com/journal/2010/1/11/scaling-down-for-tactical-apps-with-azure.html The good news is that MS has some plans under its sleeve to really improve the situation. Don’t worry too much about the pricing as it stands now, it will be irrelevant soon
Joannes Vermorel
April 6, 2010 at 07:46
Tatham raises some good points, but I agree with GP that the title is sensationalist. I know people already benefiting from the Azure platform. But these are in highly scalable scenarios, which this is not.
But rather than talk here, if you are interested in Azure (for or against) I suggest you subscribe to the ozAzure mailing list where this blog post has just popped up.
Cheers!
Steven Nagy
April 6, 2010 at 08:19
My comments are here: http://prdlxvm0001.codify.net/pipermail/ozazure/2010-April/000088.html
Briefly though, comparing a VPS to shared Ruby on Rails does not really make sense. Also, using Azure for something that could be hosted at Dodgy Hosts R Us for 5 bucks a month or whatever is kind of like using a sledge hammer to crack a nut.
David.
David Connors
April 6, 2010 at 08:43
Without getting too caught up in the debate around Tatham’s sensationalist prose (:P) I do have concerns around how Azure will build a developer community around itself.
It’s been mentioned that you a) have a great dev environment to “practice” in and b) you have some free hours per month to “try it out”.
Couple of things here: firstly, my first real go at an Azure project had me mucking around for days getting it right in the real environment… worked fine in my dev fabric, but not in the real environment. Even on newer projects I had the same problem.
You also need to learn how to manage space (blob, drive, local, cache etc), debug and log a live instance, learn the best deployment practices + just about every other thing that you need to know on a platform (platforms are big muthas usually!)… there is a hell of a lot you cannot do in the dev environment.
But… this to me is the minor point (25 hours a month might be to get this experience).
As a tech enthusiast I get the most enjoyment out of seeing my projects actually running for real, and more importantly – being used. Could you imaging getting the Windows Phone 7 API and dev kits but never getting a phone to run it on? I’d get over it very quickly – its a bit like a game for developers… “Write Windows Phone 7 apps – the game”. I like “the real” as opposed to “the game”… I have apps that are getting quite some traffic running up in Azure and that gets my juices flowing. NB Okay, okay, so most of the time development is a game to me… it’s my hobby!
One app uses Silverlight to load a DeepZoom from Azure Blob storage. The system communicates to the web instance using WCF RIA Services which then uses Linq2SQL to communicate the a SQL Azure Instance to load the metadata for the image tiles.
The back end of the app uses compute to scan Flickr automatically for new images in groups, composes them using DeepZoomTools in Azure Drive… it then pops them in to Blob storage for consumption from Silverlight. You can preview it here: http://jak.cloudapp.net/Default.aspx?guid=764fbcd0-c15f-45f3-bda5-de3ed9081ce8. The event you see here is the DigiGirlz event run by Catherine Eibner a couple of weeks ago (http://blogs.msdn.com/ceibner/archive/2010/03/10/microsoft-australia-to-host-the-first-australian-digigirlz-event.aspx).
This is a perfect scenario for Azure, and truth be told… if I just ran this in me dev environment I’d have lost *interest* very quickly!
And there is my point *interest*… for people like we developers get interested in a platform or product, we not only need access to poke and prod it but actually use it – for real. Sure, some of us have MSDN subs but a lot don’t.
Maybe there needs to be some kind of developer community version of Azure (with SQL Azure too!). Would it really hurt the system to give away some time so devs can get in there and muck around with small instances. Enterprise will still pay (hell, they may even be “evangelised” by their Azure enabled employees!).
If that goes against all thought and logic then I agree with the split share instances to make this accessible to developers who don’t have a corporate Amex so we can haz play time too
Jordan
April 6, 2010 at 09:22
Great post, and i (unlike some people on the ozazure list) do agree with most of it.
Additionally as Azure is a new technology, lack of free hosting (long term, low scale) also limits the ability to trial something for CTO superiors as a test case – this makes it harder to bring Azure into a company that has never used it before, or is thinking of using it for a future client.
Looks like a few people are picking up on this post – if nothing more, your selection of post title should see the traffic to this post jump – and that can only be a good thing.
Doug
Doug Rathbone
April 6, 2010 at 09:50
So your saying a CTO would not budget $US84 a month to test out a technology which is I am sure less that one hour of consulting time.
jamesmcc
April 6, 2010 at 10:31
I think your argument is flawed because as pointed out if you want to learn then a local dev environment is available.
What your really saying is Azure is too expensive for a little play site … which is not microsoft current target market for Azure.
What I find a tad frustrating is that its not easy to find out what is catered for in the current instance as at today ie is MVC 2 in the instance ?
jamesmcc
April 6, 2010 at 10:28
Hi James,
Jason H had a similar comment above which I replied to with this:
Tatham Oddie
April 6, 2010 at 10:31
I think MSDN needs to host its own “23/7″ sandboxed version of Azure services?
TBH the thing that annoys me most about Azure is that the Asia Pacific CPU / bandwidth costs are THREE TIMES that of USA and Europe!?!
Kenny Mac
April 6, 2010 at 12:00
Wow, you’ve just described how I’ve been feeling all weekend. Very good post and I could not agree more. . Good to know that the celebrity devs are making enough noise/standing up…hopefully they’re listening.
For now, I think I’ll just go play elsewhere
Rasel Jabbar
April 6, 2010 at 16:33
Tatham,
I agree with your points.. but a little more diplomacy next time.
Dont forget about the hundreds of dam’s, thousands of employees, millions of PCs and billions of people riding on the future of Microsoft cloud computing infrastructure.
Good post tho dude.
Jeremy
April 8, 2010 at 22:27
Tatham,
You’ve hit my sentiments precisely! Personally I would love to start playing with Azure and host very small, very low bandwidth apps to get a feel for the platform, but I can’t justify the cost.
Darren Oster
April 20, 2010 at 14:27
[...] http://blog.tatham.oddie.com.au/2010/04/05/why-windows-azure-is-not-worth-investing-in/ Exhibit 4 (click to enlarge): http://www.microsoft.com/windowsazure/offers/popup.aspx?lang=en&locale=en-US&offer=MS-AZR-0005P a/o 4/20/2010 Exhibit 5 (click to enlarge): http://www.microsoft.com/windowsazure/pricing/ a/o 4/20/2010 [...]
Windows Azure (MSFT ONLINE charge) Pricing Model Causes Great Confusion and Financial Repercussions to Developers | Charles Digital Life – Tech
April 21, 2010 at 10:16
I agree with you. I hope MS will rollback their actual price plan and will publish new one.
IMHO three point should be mandatory for a better plan:
– a free and small option (like Google App Engine)
– an automatic limiting system. I want my app stops working if I reach my limit. Today it continues to run.
– no credit card required… please!
fabrizio
Fabrizio
July 9, 2010 at 19:44
Did you see the Azure Appliance announcement? http://www.microsoft.com/windowsazure/appliance/
Hopefully someone will use it to make a cheaper clone of Microsoft’s Azure hosting.
russ
July 14, 2010 at 07:20
C’mon, folks. Microsoft makes MONEY, not software. Who cares as long as it makes them another billion?
ricky
March 25, 2011 at 16:36
Azure looks really boring anyway
Baxter
May 24, 2011 at 20:39
I would like to add to this conversation, Several years ago I developed a small work order service application in .NET, I always have difficulty selling to really small companies because of the SQL Server requirement. I decided to try Azure SQL as a test to see if I could connect my current application to there service. I signed up for a trial account, ran the sql scripts to create the data base. Connected my desktop application. Created a couple of work orders and invoices to see how the performance was. The performance was OK. I am the only person who has ever used this data base, it has two work orders and two invoices. Next month I get an invoice charge on my credit card for just under $90 bucks for transfer fee’s.
Dont fucking think I will ever use there service as it’s pricing model suck ass.
Vince
August 10, 2011 at 06:51
Seen the name banded around quite a bit, but didn’t take a look until today.
The pricing was the immediate flag for me. I havent really looked into any cloud based solutions but my understanding is that their basic draw is their on demand scaling.
The problem is that when something scales upward it goes from something small to something big, and yet, the pricing is prohibitive to anything small. I see people purport that the next youtube, twitter, facebooks can emerge from such platforms, yet these companies didn’t monetise for years. The costs required would be equal to or greater than dedicated platform alternatives (and at least you would have equipment to sell if it all went south).
I’m not even sure this is appropriate for many big boys, can you imagine what twitters bill would be for a month of using this?
It may be worth also pointing out that scaling dedicated platforms isn’t difficult. Costly yes, but they’d still have all that money they didn’t spend on database requests.
afly
December 12, 2011 at 20:36