Hate on Microsoft all you want but this is a great thing. If Microsoft feels they want to play in the infrastructure as a service game they need to earn the customer's trust.
Price matching is certainly one way to do it. Azure does run linux by the way (a lot of people don't know that).
One thing I've been going back and forth about is whether I "trust" Google App Engine or Microsoft Azure more. Both are juggernauts but Microsoft seems very intent on staying and competing in this space.
For what it's worth I've used Azure before and don't find the interface too bad. Also for spinning up the latest and greatest Windows Server instance you can't really beat Azure.
>"Hate on Microsoft all you want but this is a great thing. If Microsoft feels they want to play in the infrastructure as a service game they need to earn the customer's trust."
Interestingly, the folks I deal with some folks who run some large-scale Microsoft services for a living want nothing to do with Azure in its current state and use AWS instead.
Apparently, Azure lacks some pretty basic functionality like the ability to bind multiple addresses to an instance [0] and makes running legacy versions of Windows a pain - deal breakers for production and test purposes.
On the other hand Azure has some awesome features not available in AWS. For example you can upgrade an Azure instance to a bigger size with zero downtime.
Out of curiosity, why would you need to bind an arbitrary IP to a VM? What's wrong with the one it's presumably got?
I know some places can move IPs around at will, but it seems like the wrong layer in the stack to do it. Aren't IPs something you want to assign based on your network topology? If you need an arbitrary identifier that can move between machines, what's wrong with DNS? (And how are you not re-inventing DNS?) (and the names are more memorable)
If you do have the capability to do this, does it not tie you to that provider? Presumably the IPs would be allocated to the provider, so switching to a new provider would necessitate a new IP, whereas DNS addresses can point anywhere.
Unless I'm completely misunderstanding (I've never tried these shenanigans with IPs), so please correct me if I am.
Some DNS clients ignore the TTL, or have their own minimums.
One of the worst offenders, ironically, is Microsoft's own Windows XP which is notorious for having a minimum TTL of 10 minutes, forcing people away from DNS as a means of failover.
Strikes me that this is intentional, not some thing that they are structurally incapable of doing - it is not as if there is a massive engineering effort for example, to support such a thing.
Never the less, I am in the same boat as your friends in many cases. In my consulting practice I do lots of SQL Server tuning work and have found very few cases where I can tell a client that going to Azure would be a net benefit for the database. Even with the new dedicated SQL instances, on-premise will almost always get you more perf per dollar. (though the Azure stuff does come with less maintenance headaches in theory)
It gets even better than that if you're in an approved accelerator / incubator: you get $60,000 USD in hosting credits, at a rate of $5,000 a month. We're entirely hosted on Azure using their Linux VMs and have had a great experience with them so far.
I think the reasoning is that if you convince an accelerator/incubator to accept you your product has a higher chance of succeeding. This will mean that the chance of you still existing (and starting to pay) after 3 years should be significantly higher.
Many companies have discounts that they can afford to give, but they make it a bit of a hassle to get them, so the bulk of customers keep paying full price but they can also retain the miserly customers at a reduced profit margin.
For example, some food retailers publish discount coupons in newspapers, so the discount is only available to people willing to go through the hassle.
It's also worth pointing out that BizSpark imposes no requirements or obligations on the startup. You are free to join BizSpark and never use any of its services and you don't owe MS anything at all.
It is also worth point out you can add up to 8 developer accounts (now seems to be 4?) under bizspark programme and the $150 credit is per developer, so the total saving for 3 yrs is up to $43200 (crazy?)
Also, you can enjoy additional discounted hourly rate for up to 33%, aws/google can't compete with that.
+1 to that
Bizspark is an awesome program for startups. We are running linux on it w/o any issues
[~3 servers come free]
Best part - there are no restrictions on what s/w stack you run..
How good is the support? I ask because last time I tried HyperV I had lots of really lame issues. Like breaking the CD-ROM drive access and their workaround was to copy over an ISO and mount that locally or something like that. It felt like Linux was an unfinished offering. Yeah, I mean it ran... but not well. I figure some of what I ran into was user error since its a new tech to me but I've talked to a few people who have had serious HyperV issues so I'm not going to take all the blame.
I bring this up because I'd imagine those Linux VMs are sitting on top HyperV. I'll be very happy to hear that my fears are unfounded and they run great though.
I'm a Linux guy (well, I was a year ago :), but I'm working for Microsoft now, and get $150/mo free on Azure ; although I haven't used many of the advanced features, I'm running a few tiny instance, both Windows and Linux; it works great and has been pretty reliable (and the UI is nicer than AWS :)
The company I work for tried hard to make HyperV work, but they ended up ditching it as an inferior product.
It definitely wasn't up to the enterprise standards they were expecting. I don't know what were the specific issues they ran into, though (I'm on the development side currently).
While I've never used a Linux VM on Azure, I can say that every question I've had was answered quickly and extremely thoroughly. I've never ran into anything that support was unable to help with.
Azure to App Engine isn't apples to apples. For example, you could write an app to run on Azure and be able to change platforms to chase better prices or capabilities later, rather than having to use App Engine at inflated prices forever after.
<rant class="disgruntled">
I've used Azure's cloud hosting in production for a half dozen sites and I have to say it's driving me nuts.
We tried to flip DNS to Azure and it wouldn't resolve. It took our senior developer two days of talking to Azure support to solve. Ended up being their fault.
Today I ported some work using PHP's mail() function, which evidently Azure doesn't support?
This is probably our own fault for switching environments but we've moved sites from Media Temple and GoDaddy to Azure and consistently had problems.
Needless to say, a price drop isn't making me any more excited about Azure. Their UI is convoluted (though beautiful, for sure) and I've had nothing but issues with it.
</rant>
If you (or anyone in the this thread) would be so kind as to write up a description of some specific functionality or interactions that we can improve and email them to me (m/a/r/a/y at microsoft dotcom) I would appreciate it. I can't promise instant gratification results, but I will do my best to get your feedback to the right person.
I can give some point here:
(A cluster below is hadoop cluster)
- Machines reboot without any notifications.
- Blob storage gets throttled ( i was told by a support staff) and at times become unreachable.
- Some clusters i create, go into "Operational Mode" in that beautiful UI. and there is no way to delete that cluster!
- At one instance, a cluster failed because, i was told, some internal azure tables got deleted!
- Start up time (for PasS) for a cluster is 13+ mins regardless of how many machines you have in it.
I've had some issues with Azure too and have fought it a bit. My app is Node/Postgres based. I'm pretty positive if I had developed a C#/SQL Server app, then Azure would be almost entirely smooth sailing. I can't really blame MS for targeting their own tech first. Node, PHP and the like are not quite as first class citizens in Azure yet.
I never had trouble with any .net stack stuff but when i tried node I had some issues. Ultimately I used the MS tooling to provision the site the first (webmatrix) and from then on everything was fine.
I'm guessing that they try to manage how email is sent from their infrastructure in an attempt to curb spam and prevent their IP ranges from being blacklisted and blocked.
Well that's disappointing then. I wasn't particularly impressed with the service I got at my various hosts' homes in the UK. Is it really the case that the last mile physical infrastructure is considered common there so any provider can use it?
Yes - the UK telephone network is regulated, and any provider can supply services over the formerly-public BT network. ADSL2+ is available up to ~20Mbps; there's also BT's Infinity, which I think is FTTC with VDSL - that's up to about 70Mbps.
Virgin Media has a private cable network, which is great if you're not in an oversubscribed area - bandwidth up to 152Mbps is available for low cost.
In general the situation is pretty good, with a pretty wide variety of options.
The UK isn't really a "European" country in this regard; it still prefers inefficient private quasi-monopolies to properly functioning state services.
The UK "last mile" infrastructure is unbundled so you can get your DSL service through a wide variety of providers - but actual line speed upgrades and faults are still at the mercy of BT incompetence, and everyone has to pay them £15/month for "line rental" even if you don't want POTS.
That was a bit of a troll, yes; I'm just about old enough to remember six week lead times for phone lines, slam-door trains, but not quite the three-day week.
My argument is that state services don't have to be awful, as a lot of people seem to think. It depends on local political culture. British culture has American pro-privatisation far more than the German/Nordic tradition of efficient services.
Estonia is doing rather well at state-supported internet.
Sky. If you argue with them you'll get a good deal. I don't have Sky TV either.
Edit: you need a provider that can afford to make a loss for the sake of customer retention. Sky is a primary example as their broadband is a tiny portion of their business.
I would call it Central Europe, but yeah. Fiber to home 6 years ago, now it's 250Mbit for 15EUR. And it's not sole company either. The cities I live(d) in (in Czech Republic/Slovakia) have multiple ISPs offering 100+Mbit services. Coverage is great as well (in cities).
Google is hopefully going to fill the fiber internet gap. It's interesting to see Time Warner and AT&T scramble to put out their own "gigabit" services, after Google announces coming to Austin.
Microsoft created an entirely new instance type to match EC2 prices, ostensibly because EC2 instances don't include load balancing.
And the performance of the new "basic" instances will be reduced:
Basic instances will have similar performance characteristics
to AWS’s equivalent instances while the Standard instances
will maintain their favorable performance.
Anyone know how "favorable" Azure performance is on their standard instance type?
I still don't know why people find these "low" prices exciting. A most minimum spec VM in cloud still costs $380 per year. Considering these providers have advantage of scale plus load balancing ability, the cost should be 1/10th of this. I was looking for new host for my site and after I looked at Azure/AWS, I was floored how expensive these things are. Cloud prices had been riding in clouds for quite sometime. It has now came down but it's still way up there. The low margin claims by cloud providers is pure bullshit. The fact is that all of these people could have lowered their prices for quite sometime but they just chugged along until there was competitive pressure. Eventually, someone will become monopolist in this area and they would have free ride and all of these guys are in game to become that monopoly.
A most minimum spec VM in cloud still costs $380 per year.
Not sure where you are getting that. On Azure, it is $1612=$192/year[1]. If your "in cloud" means other large providers as well, Digital Ocean will do it for $512=$60/year[2].
If you are prepared to go with smaller providers, then a very quick look around finds providers at $5/year[3].
I've worked on pricing at a large (enterprise) hosting company and I can assure you the "low margin" claim isn't "pure bullshit".
The fact is that all of these people could have lowered their prices for quite sometime but they just chugged along until there was competitive pressure. Eventually, someone will become monopolist in this area and they would have free ride and all of these guys are in game to become that monopoly.
Don't these two claims conflict with each other? Prices are dropping because of competitive pressure. If someone will become monopolist in this area AND margins are as high as you seem to think, won't it be simple for another provider to enter the market with slightly lower prices and take market share from the incumbent monopolist?
No low margins at all. Don't forget the bandwidth pricing. No low margins there at all. Digital Ocean includes up to 5 TB of free transfer. That would cost $614 per month at AWS. Not to mention dedicated server providers like OVH where you can get a modern Xeon and unmetered 300 Mbps bandwidth for $ 137 per month.
I've worked on CDN pricing, too, although mostly outside the US market.
Depending on who you are, in some circumstances you can actually make money from cross connect charges. This is typically the case if you are a telco, but it depends on your place in the market.
You can see some hints of this in the differing ingress and egress pricing on AWS.
Anyway: Yes, AWS bandwidth is expensive and most people would be better off going elsewhere if you are using a lot. But don't be too quick to judge the margins because we have no way of knowing what Amazon pays under what circumstances.
According to the Azure calculator an XS VM costs $0.02/hr, which comes to about $175 a year. That is less than an 2 hours worth of sysadmin cost, excluding the hardware costs.
I think Azure's biggest hurdle so far is the association with Microsoft and people assuming "well it must be expensive" only to be proven right when they sign up.
That being said, some parts of Azure are pretty cheap and for small site it's affordable as it is now. You can get 10 free websites and a couple decent sized databases for around $40 a month which gives you better service and scalability than you would get from a hosting provider at $40 a month.
If you're running a business or have some real traffic though, expect to pay considerably more for the service. It can really expensive, very fast.
Yes, definitely. This is a really annoying drawback of Azure websites, because apart from that it is a really good little service for lighter loads. We've found great success in using it for API services and the like that get little traffic, but could do with being separated out from the core website.
It's actually worse than it just being $39/month, because you can't use the free or shared tiers, which is even more annoying.
Azure peeps: please change this! It makes it very hard to justify using Azure websites when you could just get your own VM and install an SSL cert on that with no restrictions for lot less.
We spend about £1500 on Azure a month. It's considerably cheaper than a quarter rack with a ton of bandwidth and all the associated license management. It's cheap but not budget.
I don't understand why Microsoft doesn't drastically reduce the price of their Windows instances.
Microsoft is matching all of AWS prices, which means the Windows servers are more expensive than the Linux servers. Why not set your Windows server prices as the same as your Linux servers since you don't have to pay the licensing to yourself. This would make azure the easy choice for those running Windows servers.
Maybe I'm missing some anti-competition law here or something, but it sounds funny that Microsoft is price matching Amazon's prices on Microsoft Windows Servers.
They do have to pay licensing, its money that goes from one business unit to another within the Microsoft Corp, thus to not destroy the Azure budget, they pass the license costs to you instead of eating it.
It's not as if one business unit can't give a "90% discount" to the other if the grand common leadership thinks it's a good idea... now, that would be anticompetitive to all other hosting providers, but it would makes sense as a tactic in the "great war against *nix" that they are 100% loosing on the server side :)
Azure uses HyperV[1] as their virtualization technology. That isn't the same as Windows at all, although it is included in Windows Server.
While I have had more reboots on Azure than on AWS, it isn't weekly by any means - more like every 3 months or so. It hasn't been often enough for it to be a problem (low usage environment though, so YMMV).
Yes, Azure VMs run on top of Hyper-V. Which means that the physical machine must reboot after every Hyper-V update.
However, each physical machine also runs an Azure Host Agent, which runs on top of Windows Server Core. While the use of Server Core helps cut down on reboots, it doesn't eliminate them.
But then, why doesn't Hyper-V just stay up while it reboots the VM containing the Host Agent? Because the virtual disk drivers are implemented for Windows Server. (Hyper-V has the concept of a management OS, which has privileged access to the hardware.)
As the underdog competing against ESX, Microsoft had to make it easy for businesses to adopt Hyper-V.
By hooking up Hyper-V to the Windows management OS, they get to leverage Windows device drivers. And this, in turn, ensures broader hardware compatibility.
Of course, this goal became unnecessary for Azure, where Microsoft controls the hardware. But Hyper-V had already been designed that way.
Certainly, Server Core is pretty stripped-down. The Russinovich blog says they reboot about once a month, but the Azure support blog says they reboot every 2-3 months.
"While this type of proactive notification of planned maintenance is not currently provided, we encourage you to provide comments on this topic so we can take the feedback to the product teams."
In my experience (running a Windows vm) the forced reboot is more like every two months. Also they have a feature called availability sets where you can load balance two VMs and they guarantee that one of them will always be up.
Edit: Looking back at my email history, it's much less often than that. The interval between the previous two reboots was 6 months. They also gave about two weeks notice when it happened.
But in order to use Availability Sets, you MUST use their load-balancer thingy and share a single IP address. Meaning if you want to individually address each server, you've gotta get creative with port mapping. It's very annoying.
I just had this happen recently and there was no warning. Perhaps it depends on the "level" of services you are using. My instance was just an "extra small" vm.
Back when we were using Azure heavily, the answer from the Azure team members I was talking to was that they had no mechanism to warn me when the machine was going to go down.
The primary workload we were running was video game servers. The server stores all state ephemerally in RAM tied to the process. When it goes down, that play session is over and all players are booted. This was just a constraint of the software that we had to live with.
What I was looking for was some kind of heads up so that I could at least try and take down the process gracefully, but that wasn't possible.
As said, this was a while ago. Hopefully the situation is now different.
I'm not sure what you are saying is completely correct. However in the cloud you should design for that, everything will eventually fail and your service should manage. ie. avoid stateful VMs, provide sufficient redundancy etc. Design for failure. This is not just about Azure, it's a common cloud design practice.
Yes, you absolutely _should_ design for that. That's why it's kind of hard to get mad at this shortcoming, because the answer is just that you should have been following best practices anyway.
In practice, though, I often found myself getting annoyed that I had that enforced on me. On AWS or most other providers I've used, if you have a workload that isn't particularly mission critical you can get away with just expecting the machine to always stay up. It'll probably only go down once a year or so. Obviously not the best way to design a system, but a handy tradeoff to be able to make in order to save a little cash or complexity.
Well, the world does not move on ideal conditions. I have used AWS and Azure extensively. My use case has been mainly around hadoop ecosystem. Here is what happens on azure(HDInsights Specifically):
You have a cluster running, that does some work at certain times in a day. Out of the blue, Several machine would be "re-imaged" and that means it would delete all the logs, working directories etc.Its like a snapshot restore. So when this happens, you have lost any custom changes you had done on the cluster. You lose any 3rd party jars you had placed.
Those machines, join back your cluster and your jobs would no longer run because of missing jars Unless you program to download all files on every reboot.
Most important thing though is lack of prior notifications. I have had cases when a "re-image" happend when my CPU was 100% utilised and my RAM was used 90%. So my box was working on something and azure decided to apply a patch!!
As per frequency, its once every two weeks, sometimes more.
I don't really understand what the difference is between Linux running on a Linux host and Linux running on a Windows host. You're going to require updates and patches for both.
I haven't noticed any serious downtime on Azure for this reason. Probably 1 reboot in the last few months, with plenty of notice (and a fairly specific timespan).
It's monthly, however even that isn't really acceptable. So far I've been too lazy to figure out if one can setup the live migration stuff to act as workaround for this...not sure. It's definitely something they should at least address / give an official workaround for, or preferably just transparently handle it.
It's worth noting that regardless of the fact that the underlying platform supports live migration, providing it as a service is requires an additional effort.
If users will favour GCE because of reduced downtime support due to the live migration support (or simply favour GCE because of they perceive it as a more advanced platform), then other providers will follow.
Right now it seems that it's all a price war, not yet a feature war. But prices can go down only to a point, then feature war will start showing it's effects, and it will be interesting.
It's interesting that MSFT has already reduced Azure's price early this January to counter the price reductions by AWS back in January. And also in April 2013, MSFT responded AWS's price drop with 21%-33% Azure price reduction.
Are we in the era of cloud price-bidding war? And the ultimate beneficiary are us the users.
We have to hope they never form a sort of cartel or join an informal alliance. Until now we are really getting a lot of benefits due to this competition
I can't help but think that they already were in a sort of informal alliance.
It took Google dropping prices for both AWS and Azure to drop... clearly they were both skimming profits off the top for the last while.
Of course, a company does need to profit and I'm not accusing either of them of doing anything wrong, I'm just trying to say that these sorts of alliances form accidentally and naturally; both companies hit a competitive price and then hardware gets cheaper; software gets better... they could drop prices, but why cut into profits when there's no need? Inertia is a powerful force. It takes something actually happening to kick those price drops into action, generally.
> these sorts of alliances form accidentally and naturally
The airline-ticket market is kind of fascinating to watch from that perspective. On directly competing routes major carriers often end up with the same fares and matching each others' moves, but the exact process of who leads and who follows is complex. There is all kinds of informal signalling where companies attempt to figure out the others' intentions without directly colluding. For example, a company might trial a $50 fare increase in response to increasing costs, and hope others will match it. If others don't, they might be forced to roll back the increase to avoid losing market share, admitting the trial was unsuccessful. It's in effect an offer of, "hey everyone, I think we should raise prices $50 on this route, do you match?", followed by "ok, I guess not, retracting" if they don't, but not explicit.
If that were to happen there is nothing stopping new market entrants from quickly gaining market share simply with lower prices.
DigitalOcean has done this to some extent (although it is unclear how much of their pricing relies on venture funding and generous payment terms for hardware/data centers space combined with rapid growth.)
They really need to cut prices. I created a card-lookup site for a collectible card game (http://www.lotrlcg.com/).
Originally it was written in PHP and was on regular shared hosting through InMotion Hosting for about $5/month.
I rewrote it in ASP MCV and hosted it through Azure and they are charging about $20/month - a jump that I was not expecting. The convenience of being fully integrated into the MS stack is great, but I was planning on moving everything with project I start. Hopefully the price difference will be good enough to stay.
I lost count of the times they mentioned Amazon or AWS in this article. This is going to be exciting, Microsoft are really gunning for them.
Interestingly it looked like the price comparison tables were based on on-demand pricing. Does azure have something similar to AWS's reserved instance model?
I think it's more to do about branding consistency. Azure is far more than just 'Windows' now, you have: Media Services, CDN, Mobile Services, Cache, etc.
This is a great strategy by Microsoft and it's part of "1st Mobile & 1st Cloud" strategy which is being adopted by Microsoft New CEO. As Nokia X came into market where user will get 7 GB of free cloud & this will increased upto 10 GB if user allow camera access to the cloud. Cut-throat competition is about to begin in cloud market, each and every player are ready to compete with attractive offering.
Does anyone else feel like Rackspace is just getting left behind in the IaaS business? Their prices and services offered is getting out of whack compared to everyone else.
I really wish they would have included some price cuts on the web site services.. Or at least give us an extra small instance for standard web sites. Scott Hanselman said it was "coming soon" back in August: http://www.hanselman.com/blog/PennyPinchingInTheCloudWhenDoA...
This is a good thing, but I wonder how this will affect people like Digital Ocean, Rackspace, etc.
Still, this is a really great thing. Competing on price and/or features will only improve the environment for everyone. Probably not quite a race to the bottom, but there are deep pockets involved in the AMZN/GOOG/MSFT collective.
I expect all the management tools to start to get a lot better.
Seems like someone at Microsoft is reading this thread. In that case, an incubator I'm involved with (Tandem Entrepreneurs in the bay area) would like to get that sweet incubator deal if it's still on. We're already on a similar deal with AWS.
And ahem, Google cloud folks if you're reading this... you know what to do.
does anyone have a link to the usability of AWS v Google vs Azure? I'm looking to setup a personal server for all kinds of data wrangling, personal projects, websites etc. Im wondering what the most cost effective solution might be.
I feel like this is one of the few times where I've seen capitalism work clearly and unambiguously along the lines that Adam Smith had in mind. I'm deriving something approaching mischievous glee.
I don't know, I see it every week a block from my house. Chevron drops the price of gas twenty cents and before the end of the day the citgo across the street has dropped theirs 21 cents.
Of course when one raises prices, the others follow right back up.
Steve Martin comes off as oddly defensive when announcing these changes
I’ve yet to meet a developer who travels just to hear about pricing updates,
so you won’t see us take the stage at Build and use the opening moment
to announce a price cut. This isn’t retail; innovation matters.
I’ll cover pricing right here, right now.
I'm not sure what purpose it serves except for making a stab at Amazon and Google.
That's a very cynical position to take. For once, here is a large market player focussing on providing delivering value via the product and not wasting time on pricing. In many cases, even with these price reductions (and their claim to match AWS/GCE), pricing is not the driver for hosting adoption - rather features, ease of use, performance etc. It is refreshing to see one of the big boys elevate this.
It does set a high benchmark for what they are presenting: saying they have big innovation to announce ... and then not deliver ...
What he means is: "we weren't expecting the Google and Amazon price cuts, so we aren't going to change our carefully planned keynote to announce this."
Ah, image tables, why use <table> when you can just add an image right? And I bet these tables were drawn in Paint. You can clearly see that in the first table the borders don't even match (bottom left).
I am a Microsoft fan, but they are doing the cloud all wrong.
To setup Azure, you have to use silverlight.
To access the servers, you use remote desktop.
Microsoft office 365 is $100 PER YEAR - much more than I would pay for the full suite (over 4years of average use)
If you want to use SQL server, expect to pay $$$ - and integration or reporting services are not included.
There is another feature that is actually nice: if allocate one server, you get 5 extra services automatically, including active directory, block storage, and network setup. if you do not know each part of the cloud, this helps you out, but it is done the "Microsoft way" and can be confusing compared to other cloud services.
Seems to have been a more general change; the Bing Webmaster Tools portal dropped Silverlight about a year ago as well. Not sure if there was an official announcement or they've just quietly decided to move away from it.
AFAIK Silverlight is only required admin the legacy share cache service, which has been end-of-lifed (hence the reluctance to spend the effort needed to support it in the web-based portal). They recently announced that they would keep the Silverlight legacy portal around for a bit longer do give people more time to move away from the shared cache service.
I use Azure every day and I haven't needed to touch the Silverlight portal in months.
Price matching is certainly one way to do it. Azure does run linux by the way (a lot of people don't know that).
One thing I've been going back and forth about is whether I "trust" Google App Engine or Microsoft Azure more. Both are juggernauts but Microsoft seems very intent on staying and competing in this space.
For what it's worth I've used Azure before and don't find the interface too bad. Also for spinning up the latest and greatest Windows Server instance you can't really beat Azure.