CloudEnterprise.info

Archive for the ‘Commentary’ Category

Don’t put all eggs in one basket! June 14, 2012 outage at Amazon Web Services affected many customers and other clouds that rely on AWS (e.g. Heroku). Instead of going back to the “is cloud reliable” debate, we need to acknowledge that no single service will ever be 100% reliable, and only real solution is using more that one service provider.

Remember when Apple launched their iCloud service last year? Remember what made them architecturally so different from 99.9% of other “cloud” services out there? They used both Amazon Web Services and Microsoft Windows Azure as the underlying platforms. Does anyone care to guess why? Perhaps the answer in the latest news. Just read these articles about Amazon’s or Azure’s outages. Can you find iCloud mentioned anywhere as one of the affected services? No. You will see Heroku, Quora, Parse, and Pinterest – but not Apple. If one cloud fails – they still have the other one to use.

I work for a cloud platform myself (Jelastic PaaS) – and let me make it clear: no matter how much work we put into making it as reliable as we can – any services can (and will eventually) have an outage. Even a service with multiple availability zones (like AWS or Azure) will fail from time to time (happened already). Don’t cheat yourself – if you need real redundancy – use more than one provider, and do yourself a favor – check their backend platform. If you think that using AWS and Heroku is redundant – you are wrong – they are both running on AWS.

And yes, this means that you need to try to pick the services that accept the same application code. If one of the services requires your application re-written your development cost will double (e.g services like Google App Engine require pretty much complete application re-write to use it – a bad choice as a second platform.)

This is one of the reasons why in Jelastic we made a couple of important design decisions early on: make it available from multiple completely independent hosters (not Amazon, but actual real credible hosting companies) and make it 100% code compatible for any Java applications (no APIs to code to, no code changes required).

Don’t want to be in the next outage news? Pick 2 hosters and get yourself some piece of mind (obviously do check your failover to make sure you can safely stop your service at one and switch to the other one! – the only redundancy that works is the one that you test as often as you can.)

Advertisements

Everyone suspected that Windows Azure was not a blazing success. Now the official stats seem to suggest that it is pretty much a failure.

Microsoft does not tell the exact number of users it has on their cloud platform – Windows Azure, however a couple of day ago we got the latest vague estimate from the company. Quoting Mary Jo:

“On May 8, Microsoft Corporate Vice President of Azure Marketing Bob Kelly provided Merrill Lynch Technology Conference attendees with another tally tidbit. Kelly said Microsoft now has “high tens of thousands of customers” for Windows Azure.”

That was it, so now let me try to interpret what we have heard:

  • I assume that “high tens of thousands” means 50,000 to 100,000 customers,
  • Is that for paying customers or does that include free ones? I assume that if they were paying – Bob would have told us so,
  • Does that include Microsoft’s own teams? Probably, yes. Microsoft teams traditionally view other teams within the company as “customers”,
  • If I have a team of developers working on a project – do these get counted as 1 customer or multiple customers? If everyone is counted the number would have to be divided further. I would assume that the answer is 1 – otherwise, considering that SaaS development and operations are team efforts, the resulting number would get too small.

Now, let me be straight on that, if these assumptions are right, this is a very small number for a platform that was publicly launched in October 2008.

Just to put that in perspective, the company for which I work now – Jelastic (Java PaaS) – launched public beta in October 2011 and last month announced 15,000 signed-up users (including free, trial and beta).

As much as I love our marketing team, the marketing resources that we have are minuscule compared to that of Microsoft, so considering all the efforts that Microsoft made touting Azure everywhere and all the .NET developers their marketing can reach – Bob Kelly’s number is incredibly low.

The previous datapoints that Mary Jo quotes are in line with the current number:

“In 2010, the Redmondians said they had 10,000 Azure customers. In 2011, it was 31,000. (Microsoft officials declined to say if any of these were Microsoft users and how many were paying customers.)”

I have a lot of friends at Microsoft and a lot of sympathy toward the company, so I really hope that some of the assumptions that I made are wrong. If so, I think Microsoft should be more transparent about the way they count “customers”. Giving the number and then letting everyone make their best guesses on how it was counted – is a very bad tactics. People just assume the worst case scenario and this damages, rather than improves the company image.

 

supermarket-vs-farmThanks to Nicholas Carr everyone loves to talk about Cloud Computing as electricity and how we are transitioning from own power-stations to central grid. After a lot of discussions about “the cloud”, it occurred to me recently that there is a far better model to make cloud idea easy to understand – and it is supermarket.

Electricity is kind of a wrong model. Not because so many folks are now trying to get the pendulum swing back and get to solar panels and other micro-generators, but also because software and IT services are much less uniform than electrical current. There’s no single “IT utility current” you can get from your network outlet to solve all IT needs.

Supermarkets seem a far closer paradigm. There was a period of natural household economy when basically people were raising their own crops and more or less producing most of the stuff they needed day to day. And of course we do not really do this anymore because – as Nicholas so eloquently demonstrated in his books – specialization and mass production make more economic sense.

All the talks about security, privacy, lack of control and so all totally apply here. When you grow your own potatoes you can be by far more sure that no pesticides get in there. However, we just don’t do this anymore apart from maybe a fraction of people growing some plants for fun and personal joy.

Even more, we now have the whole new segment in agriculture – organic/bio food – which charges premium for real or perceived additional quality and safety (depending on the country there might or might not be real certifications and controls involved).

Seems to me that this is exactly the direction in which we are heading. Maybe with the difference that in IT, the quality of service might actually be easier to track than in the food industry – so we are probably in a better shape than anyone else.

Are online services ever going to be 100% secure? If not should the insurance industry kick in?

A few days ago Google Apps had an issue with some Google Docs became accessible to other Google users beyond the security set on the docs. To quote from Google:

As we noted in the Google Docs Help Forum yesterday, we’ve identified and fixed a bug where a very small percentage of users shared some of their documents inadvertently. The inadvertent sharing was limited to people with whom the document owner, or a collaborator with sharing rights, had previously shared a document… We believe the issue affected less than 0.05% of all documents…

This obviously is not fun, and 0.05% can be a pretty big number of documents and who knows how these got spread across customers. However, what I wonder is whether this is actually an insurance industry rather than just technology opportunity.

Seriously, you install fire alarms, etc. in your house but you probably still insure it against fire (and not, say, live in the middle of a field because houses can burn). Does this make sense?

Tags: , , , ,

Richard Stallman is surprisingly wrong in his interview to The Guardian. In the interview he seems to be confusing the cloud computing/SaaS technology and the current lock-in way big vendors are using it.

Let me quote the interview:

Cloud computing – where IT power is delivered over the internet as you need it, rather than drawn from a desktop computer – has gained currency in recent years. Large internet and technology companies including Google, Microsoft and Amazon are pushing forward their plans to deliver information and software over the net.

But Richard Stallman, founder of the Free Software Foundation and creator of the computer operating system GNU, said that cloud computing was simply a trap aimed at forcing more people to buy into locked, proprietary systems that would cost them more and more over time.

The logical mistake that the article makes is confusing the technology and the implementation. There is nothing inherent in cloud computing approach that creates a vendor lock-in more than on-premise systems. After all, we have heard multiple times Microsoft being accused of creating lock-in with their Office file-formats or undocumented internal protocols.

Cloud computing is fundamentally about outsourcing non-core IT infrastructure such as messaging to others who can run it cost-effectively for you in their datacenters enjoying the economy of scale and opportunity to perfect this one system they care about. Because of this, the approach currently has inherent economic advantages.

The fact that current implementations mostly try to lock customer data in providers’ datacenters is to be expected in any early market. This is a problem which is likely will need to be addressed due to valid concerns Richard is expressing. I believe that this will happen – a this is happening in the on-premise world partly due to the anti-trust governmental activities, partly thanks to the open source movement and overall change to the popular opinion.

Technorati:


RSS My company’s main blog

My Recent Tweets

Blogroll

Legal

The posts on this blog are provided “as is” with no warranties and confer no rights. The opinions expressed on this site are mine and mine alone, and do not necessarily represent those of my employer Jelastic or anyone else for that matter. All trademarks acknowledged.

© 2008-2012 Dmitry Sotnikov

%d bloggers like this: