The AWS re:Invent 2013 is over and the Amazon Web Services (AWS) continue to reach out to the corporate clients with some new services. After AWS has established itself as a leading infrastructure provider and enabler for startups and new business models in the cloud, the company from Seattle tries to get one foot directly into the lucrative business environment for quite some time. Current public cloud market figures for 2017 from IDC ($107 billion) and Gartner ($244 billion) to give AWS tailwind and encourage the IaaS market leader in its pure public cloud strategy.
The new services
With Amazon WorkSpaces, Amazon AppStream, AWS CloudTrail and Amazon Kinesis, Amazon introduced some interesting new services, which in particular address enterprises.
Amazon WorkSpaces is a service which provides virtual desktops based on Microsoft Windows, to build an own virtual desktop infrastructure (VDI) within the Amazon Cloud. As basis a Windows Server 2008 R2 is used, which rolls out desktops with a Windows 7 environment. All services and applications are streamed from Amazon data centers to the corresponding devices, for what the PCoIP (PC over IP) by Teradici is used. It may be desktop PCs, laptops, smartphones or tablets. In addition, Amazon WorkSpaces can be combined with a Microsoft Active Directory, what simplifies the user management. By default, the desktops are delivered with familiar applications such as Firefox or Adobe Reader/ Flash. This can be adjusted as desired by the administrators.
With Amazon WorkSpaces Amazon enters a completely new territory in which Citrix and VMware, two absolute market players, already waiting. During VMworld in Barcelona, VMware just announced the acquisition of Desktone. VDI is basically a very exciting market segment because it redeemed the corporate IT administration tasks and reduces infrastructure costs. However, this is a very young market segment. Companies are also very careful when outsourcing their desktops as, different from the traditional on-premise terminal services, the bandwidth (network, Internet, data connection) is crucial.
Amazon AppStream is a service that serves as a central backend for graphically extensive applications. With that, the actual performance of the device on which the applications are used, should no longer play a role, since all inputs and outputs are processed within the Amazon Cloud.
Since the power of the devices is likely to be more increasingly in the future, the local power can probably be disregarded. However, for the construction of a real mobile cloud, in which all the data and information are located in the cloud and the devices are only used as consumers, the service is quite interesting. Furthermore, the combination with Amazon WorkSpaces should be considered, to provide applications on devices that serve only as thin clients and require no further local intelligence and performance.
AWS CloudTrail helps to monitor and record the AWS API calls for one or more accounts. Here, calls from the AWS Management Console, the AWS Command Line Interface (CLI), own applications or third party applications are considered. The collected data are stored either in Amazon S3 or Amazon Glacier for evaluation and can be viewed via the AWS Management Console, the AWS Command Line Interface or third-party tools. At the moment, only Amazon EC2, Amazon ECS, Amazon RDS and Amazon IAM can be monitored. Amazon CloudTrail can be used free of charge. Costs incurred for storing the data to Amazon S3 and Amazon Glacier and for Amazon SNS notifications.
AWS CloudTrial belongs, even if it is not very exciting (logging), to the most important services for enterprise customers that Amazon has released lately. The collected logs assist during compliance by allowing to record all accesses to AWS services and thus demonstrate the compliance of government regulations. It is the same with security audits, which thus allow to comprehend vulnerabilities and unauthorized or erroneous data access. Amazon is well advised to expand AWS CloudTrail as soon as possible for all the other AWS services and make them available worldwide for all regions. In particular, the Europeans will be thankful.
Amazon Kinesis is a service for real-time processing of large data streams. To this end, Kinesis is able to process data streams of any size from a variety of sources. Amazon Kinesis is controlled via the AWS Management Console by assigning and saving different data streams to an application. Due to Amazon’s massive scalability there are no capacity limitations. However, the data are automatically distributed to the global Amazon data centers. Use cases for Kinesis are the usual suspects: Financial data, social media and data from the Internet of Things/ Everything (sensors, machines, etc.).
The real benefit of Kinesis, as big data solution, is the real-time processing of data. Common standard solutions on the market process the data via batch. Means the data can never be processed direct in time and at most a few minutes later. Kinesis removes this barrier and allows new possibilities for the analysis of live data.
Challenges: Public Cloud, Complexity, Self-Service, „Lock-in“
Looking at the current AWS references, the quantity and quality is impressive. Looking more closely, the top references are still startups, non-critical workloads or completely new developments that are processed. This means that most of the existing IT systems, we are talking about, are still not located in the cloud. Besides the concerns of loss of control and compliance issues, this depends on the fact that the scale-out principle makes it to complicated for businesses to migrate their applications and systems into the AWS cloud. In the end it boils down to the fact, that they have to start from scratch, because a non-distributed developed system is not working the way it should run on a distributed cloud infrastructure – key words: scalability, high availability, multi-AZ. These are costs that should not be underestimated. This means that even the migration of a supposedly simple webshop is a challenge for companies that do not have the time and the necessary cloud knowledge to develop the webshop for the (scale-out) cloud infrastructure.
In addition, the scalability and availability of an application can only be properly realized on the AWS cloud when you stick to the services and APIs that guarantee this. Furthermore, many other infrastructure-related services are available and are constantly being published, which make life clearly easier for the developer. Thus the lock-in is preprogrammed. Although I am of the opinion that a lock-in must not be bad, as long as the provider meets the desired requirements. However, a company should consider in advance whether these services are actually needed mandatory. Virtual machines and standard workloads are relatively easy to move. For services that are very close engaged into the own application architecture, it looks quite different.
Finally. Even if some market researchers predict a golden future for the public cloud, the figures should be taken with a pinch of salt. Cloud market figures are revised downwards for years. You also have to consider in each case how these numbers are actually composed. But that is not the issue here. At the end of the day it’s about what the customer wants. At re:Invent Andy Jassy once again made clear that Amazon AWS is consistently rely on the public cloud and will not invest in own private cloud solutions. You can interpret this as arrogance and ignorance towards customers, the pure will to disruption or just self-affirmation. The fact is, even if Amazon will probably build the private cloud for the CIA, they have not the resources and knowledge by far to act as a software provider on the market. Amazon AWS is a service provider. However, with Eucalyptus they have set up a powerful ally on the private cloud side, which makes it possible to build an AWS-like cloud infrastructure in the own data center
Note: Nearly all Eucalyptus customers should also be AWS customers (source: Eucalyptus). This means conversely, that some hybrid cloud infrastructures exist between on-premise Eucalyptus infrastructures and the Amazon public cloud.
Advantages: AWS Marketplace, Ecosystem, Enabler, Innovation Driver
What is mostly ignored during the discussions about Amazon AWS and corporate customers is the AWS Marketplace. In addition, Amazon also does not advertised it too much. Compared to the cloud infrastructure, customers can use to develop their own solutions, the marketplace offers full-featured software solutions from partners (eg SAP), which can be automatically rolled out on the AWS infrastructure. The cost of using the software are charged per use (hour or month). In addition, the AWS fees for the necessary infrastructure are charged. Herein lies the real added value for companies to easily outsource their existing standard systems to the cloud and to separate from the on-premise systems.
One must therefore distinguish strictly between the use of infrastructure for in-house development and operation of ready-made solutions. Both are possible in the Amazon cloud. There is also the ecosystem of partners and system integrators which help AWS customers to develop their solutions. Because, even if AWS itself is (currently still) a pure infrastructure provider, they must equally be understood as a platform for other providers and partners who operate their businesses on it. This is also the key success and advantage over other providers in the market and will increase the long-term attractiveness of corporate customers.
In addition, Amazon is the absolute driving force for innovation in the cloud, no other cloud provider technologically is able to reach at the moment. For this purpose, it does not require re:Invent. Instead, it shows almost every month anew.
Amazon AWS is – partly – suitable for enterprise IT
Depending on the country and use case the requirements vary, Amazon has to meet. European customers are mostly cautious with the data management and store the data rather in their own country. I already met with more than one customer, who was technically confident but storing the data in Ireland was not an option. In some cases it is also the lack of ease of use. This means that a company dones’t want to (re)-develop its existing application or website for the Amazon infrastructure. Reasons are the lack of time and the knowledge to implement, what may results in a longer time to market. Both can be attributed to the complexity to achieve scalability and availability at the Amazon Web Services. After all, there are not just a few API calls. Instead, the entire architecture needs to be oriented on the AWS cloud. In Amazon’s case its about the horizontal scaling (scale-out) which makes this necessary. Instead, companies prefer vertical scaling (scale-up) to migrate the existing system 1:1 and not to start from scratch, but directly achieve success in the cloud.
However, the AWS references also show that sufficient use cases for companies exist in the public cloud in which the data storage can be considered rather uncritical, as long as the data are classified before and then stored in an encrypted way.
Analysts colleague Larry Carvalho has talked with a few AWS enterprise customers at re:Invent. One customer has implemented a hosted website on AWS for less than $7,000, for what an other system integrator wanted to charge $ 70,000. Another customer has calculated that he would pay for an on-premise business intelligence solution including maintenance about $200,000 per year. On Amazon AWS he only pays $10,000 per year. On the one hand these examples show that AWS is an enabler. However, on the other hand, that security concerns in some cases are yield to cost savings.