The public cloud jungle is dominated by two really big beasts: Amazon AWS and Microsoft Azure. But how do their IaaS/PaaS services differ, and what’s to choose between them?
As part of our ongoing research into AWS vs Azure and other public cloud IaaS and PaaS services for Computing Delta, we’ve been asking senior IT professionals about their preferences. The full end user analysis, with interviews and surveys with over 200 senior IT leaders who have purchased these services, is available to Delta subscribers; click here if you do not have access but would like to see the full report in a demo. More information, including comparisons with other vendors, is available in the Cloud IaaS/PaaS Report. But here’s a brief summary of AWS vs Azure.
AWS vs Azure – the background
Based in Seattle, USA, Amazon employs 647,000 people globally (2018) and in 2018 it achieved a market capitalisation of US $1tn on revenues of US $232.887 billion. In Q2 2019 cloud contributed to 13 per cent of its revenue, compared with 11.5 per cent the previous year. AWS continues to grow at an annual rate of 38 per cent with revenues of $25.6 billion (2018). It is the most profitable part of Amazon’s many operations.
Microsoft, also based in Washington State, has 131,000 employees and recorded annual revenues of US $125.843 billion in Q4 2019. Intelligent Cloud, which includes Azure, Windows Server, SQL Server, Visual Studio, GitHub and consulting services brought in US $38.985 billion, with cloud growing at 64 per cent. Q4 2019 was the first time in more than three years that Intelligent Cloud produced more quarterly revenue than the segments containing Office and Windows.
So, both of the big beasts are growing fast, although not at the triple-digit rates of a couple of years ago. It’s hard to compare the size of AWS and Azure on a like-for-like basis, but AWS is thought by most analysts to be the largest IaaS/PaaS cloud provider (which is, no doubt, why Microsoft bundles SaaS figures into the total).
Certainly, AWS was the first to rise to prominence, thanks to Amazon’s significant early investment. The firm launched Elastic Compute Cloud in 2006, giving Amazon a good three years′ head start over competitors. Microsoft launched Azure in 2010.
The roots of AWS are in open source, and developers have long favoured the platform. However, Azure has been catching up in almost every aspect, making choosing between them more difficult than it once was. A few years back Microsoft embraced open source, something unthinkable a decade ago, and of course it now owns GitHub, bringing it closer to the powerful ‘developer lobby’ within organisations.
But Microsoft’s real advantage is its enterprise presence. Unlike Amazon, Microsoft has a notable corporate footprint, and its SaaS services, particularly O365 and Dynamics 365, give it a shopfront to introduce customers to Azure. With engineers already well versed in Microsoft tools and interfaces, adding additional cloud services from Microsoft is often a logical next step and a simpler proposition than using a third party.
“They have this practice of making things simple and packaged, bundling things and throwing free stuff in to tempt you,” said a test manager from the transport sector.
This was the primary reason respondents chose Azure over AWS (although some were using both).
“We had to make a choice between the two, and there was better synergy with other Microsoft services,” said one respondent.
On the other hand, IT leaders perceived AWS as offering a wider range of services, particularly at the cutting edge, and also as being more mature and more responsive to unique or specialised use cases.
“AWS are the leading provider. Microsoft, whilst a significant vendor, aren’t as mature in this area,” said one respondent in a large media organisation.
Which are the most important factors when looking at a vendor for public cloud IaaS / PaaS solutions?
Source: Computing Delta; n=188.
The most important factors identified by our respondents when identifying a suitable cloud vendor are shown above. They are headed up by running costs and followed by uptime/availability, security credentials and appropriate UK / EU focus.
AWS and Azure achieve similar satisfaction ratings on all these factors. The highest scores were for uptime, in which AWS had the edge, and security – which was a dead heat. Azure was rated slightly higher for its UK / EU focus, where both achieved middling scores. Both are US vendors, of course, and as such are subject to intrusive laws such as FISA 702 and the Patriot Act, which rules them out of consideration by certain European organisations. However, neither scored particularly well on the issue of running costs and the cost of adding extra services – although AWS was thought to be slightly less expensive in general.
AWS vs Azure – pricing
Unlike with on-premises software, cloud prices for storage and compute are generally in the public domain. However, that doesn’t make cost calculations simple. For example, an AWS t3.medium EC2 Linux instance in the London region with 4 GB EBS storage costs $0.0418 per hour on demand (August 2019) whereas with Spot pricing (short duration) it was $0.0142 at this time and Reserved pricing (1 or 3-year contract) was roughly half the on-demand figure. One of the complaints about both AWS and Azure was that the pricing structure was confusing and complex, although if you can work your way around it there are significant savings to be had.
As can be seen above, listed prices for virtual machines (VMs), storage and serverless are very similar. The VM price differences can be accounted for by amount and type of storage provided, and Windows VMs are unsurprisingly cheaper on Azure. Indeed, on its web site Microsoft promises to match AWS pricing for comparable Azure services and doubtless AWS would do the same if pushed. The fierce competition between the two has exerted downward pressure on prices. Amazon’s prices used to be lower, but Microsoft has the deeper pockets and has been able to keep up with AWS in this price war.
But list prices are only a small part of the picture; they are the ‘loss leaders’ designed to tempt the customer through the door. When it comes to the cost of adding additional services or of licensing certain use cases prices can soon start to rise.
“Default configurations often insecure; costs mount quickly,” said one respondent of AWS, while a CIO at a university accused Microsoft of treating students as employees for licensing purposes, thus pushing up the running costs of its services.
AWS vs Azure – support
Cloud hyperscalers tend to take a hands-off approach, relying heavily on self-service portals’ community support, with both providing well-used developer and support forums. If further configuration assistance is likely to be required it’s worth checking the availability of local specialised partners. While both Amazon and Microsoft offer support contracts of various classes, there were complaints that ‘it can be hard to get through to a human’ using the vendors’ channels. In many cases, advanced support will be down to finding a trusted local partner.
AWS vs Azure – conclusion
In the ongoing AWS vs Azure battle, comparisons based on technical or price grounds are not easy to make and the final decision is most likely to depend on what is in place already. Organisations requiring Windows integration and those who make heavy use of Dynamics, O365 or SQL Server will likely go for Azure, whereas developer-led organisations or those at the bleeding edge of AI and microservices may prefer AWS. VMware is also closely aligned with Amazon’s cloud (VMware Cloud runs on AWS) and this could be another influencing factor for those with a preference for that vendor.
As the financial figures show, neither firm is likely to go under any time soon; both are adding new services on an almost daily basis and their clouds continue to grow rapidly. Where one vendor leads the other will soon follow, typically in short time. Both have a de facto lock-in of sorts, given that it is easier to integrate services already on the platform, so it’s really a case of pick your ecosystem and run with it. But, ultimately, for general purpose use cases, the choice is likely to be a business decision, influenced as much as anything by the quality of partners on the ground.
More information, including the capability to compare satisfaction ratings with those of other vendors, is available in the Cloud IaaS/PaaS Report.