Is anyone here actually using Azure? My teammates using it have encountered all types of problems, with bad stability problems (worse than rackspace and their bimonthly outages)
I do, and think it's a good proposition for startups.
Office 365 makes it easy to setup the organisational IT (login based on your own domain, email, storage, document collaboration, chat). Add Power BI if you need fancy (realtime) analytics and charts.
Then you create a private git repository (using the same login) for your startup product and configure continuous deployment using Visual Studio Online and Azure. Most of the Visual Studio Services are node-based now, so it's cross-platform and you can deploy to Linux too.
It's also easy to add other services if you want (Slack, Zoho, Google).
All this is managed with a single login, paid with only two bills (Office 365 and Azure are separate). The Azure bill can be lowered if you get into BizSpark.
This all saves a lot of work, a single person can create and manage a very advanced infrastructure without too much effort.
We do, and we're going into Azure native tech (Reliable Actors/Services). Azure tech is designed to overcome one of the major limitations of the cloud, which is:
> bad stability problems
The cloud is inherently unstable. If you have this problem it means that your architecture isn't ready for the cloud. This is why software developers are interested in consensus/fault-tolerance in the first place. Azure might be more unstable but at the end of the day the degree of stability is completely moot. If your software isn't designed for the cloud bad things will eventually happen. It doesn't matter if it takes hours, minutes, or days.
That is the reason we chose Azure: Reliable Actors/Services are a huge boon when it comes to writing software that correctly deals with cloud conditions because they are architectural patterns that are built for it.
This is why Netflix uses chaos monkey[1]. Even the Azure SDK ships with a chaos monkey script. If fault tolerance is not a concern of yours then cloud should be your very last concern. We tried slapping on-premise architecture on the cloud (evaluating AWS, RackSpace and Azure) and we had some degree of stability problems across all of them.
The MS Azure documentation and training makes it pretty clear about application developers having to cope with "transient faults" - they even supply libraries to cope with it!
Exactly, appropriating the transient fault application block was the first step in getting our architecture to work on the cloud. However, connections to SQL are merely one part of the puzzle.
Instances of your worker nodes might also simply vanish.
* If you only have one node this is a "hard fault." Users cannot use your application until the node is restored.
* If you have more than one it's a "soft fault." Users can often immediately retry their request.
* If your own architecture is correctly designed (many nodes with fault tolerance) it is a "transient fault." Users are oblivious to the fact that one of your 100 nodes failed - even if that specific node was servicing their request.
I'm using it, because they give me €150,- in credits each month because I'm enrolled in BizSpark.
I'm a bit torn. I like Azure, the interface is much nicer than other cloud providers in my opinion, the hosted SQL works well for my needs, being able to run for free is very nice...
I'm loving the grouped SQL billing. Basicly, instead of paying seperately for each SQL database you host, you can pool all of your databases together and they'll share a billing pool. So, if one of your customers needs less performance and the other happens to use more, there's no change in what you pay.
But I have encountered stability and speed issues, especially in the beginning. Not on the VM's themselves, mind you, just in the management side of things.
Most of these issues seem to stem from the backends, the frontend is pretty slick but it can only do so much when the response time of backends is so slow. Also, it's a bit ridiculous to get an unknown error when you try to add an endpoint to a firewall, because the previous endpoint operation (Adding a different endpoint) was still in progress.
Nowadays it seems to have stabilized and improved a lot and I'm hopeful it will continue to get better.
When I had an msdn they gave me £100 a month free or something - which was super cool. With that I was able to run a lovely minecraft server for me and a couple friends :)
I use Azure and prefer it to AWS. You can get managed Azure or AWS solutions through Rackspace too, if you don't want to rely on Rackspace natively.
Anyhow, Microsoft's revenue is exploding in this space mostly because it lumps in Office 365; not really because of Azure itself, which I think is a great product. Azure has a few short falls that are annoying...still preferable to AWS though.
AWS is still very popular with extremely popular apps which drive its revenue lines. Netflix, Tinder, etc all use AWS. However from my experience with Azure they could lower their cost if they had Azure installs, AWS likes to keep instances up and spinning which eats into your costs (and they are enormous) Azure has better auto-scaling and it shows. I can't bare to imagine how much Amazon makes off of Netflix's AWS usage, you'd think they'd want to get off of their competitor's platform...but I guess it costs a lot to migrate.
