BT

The Current Status and Predictions on Cloud Computing from Cloud Leaders

by Abel Avram on Jun 23, 2011 |

Werner Vogels from Amazon and Satya Nadella from Microsoft talked about the current status of cloud computing, while Lew Tucker from Cisco and Jon Weinman from HP made some predictions on its future at GigaOM Structure Event.

In State of the AWS Cloud, Werner Vogels, CTO at Amazon, talked about what happened with AWS during the last year without covering all feature releases, but rather mentioning key developments that are of most interest to the public.

Most of AWS’s new features are suggested by customers, and they are released immediately as they are implemented without waiting on some event to announce them, said Vogels.

AWS is not just about some services provided by Amazon, but there is an entire ecosystem of services provided by Amazon’s partners which are putting together lots of data, and new applications can combine those services according to new scenarios, added Vogels.

AWS now has 5 regions after adding Tokyo and Singapore, and it holds 359B objects, a sharp increase for last year’s 50B objects, and “we haven’t reached the knee of the hockey stick yet,” said Vogels, referring to expected AWS growth. Each day they are adding computing capacity, as much as the entire Amazon.com website needed in the year 2000.

One of the main benefits of cloud computing is the opportunity to create all sorts of new applications using new scenarios, and Vogels presented several examples of companies taking advantage of AWS. Netflix created highly available software components using Chaos Monkey. soocial.com built a reliable website on AWS using multiple availability zones, DB failover, and load balancers. Oracle and SAP have certified their software for AWS, assuring customers that their products run smoothly in Amazon’s cloud.

Vogels noted that Spot Pricing, a feature allowing customers to bid for EC2 instances, is influencing how software is built because architects start considering the price for using a computing resource, and adapt their design accordingly. Vogels mentioned four types of architectures making use of Spot: map-reduce, data grid software, queues, and check point restore software. All these can stop processing when the price of EC2 resources rise, and can restart when the price goes back down. Financial or video processing companies are the main users of this feature, and they employ “fascinating bidding techniques” to lower the computational price. Vogels announced the possibility to now use Elastic Map-Reduce along with Spot Pricing.

In Microsoft and the Cloud: What's Next?, Eric Savitz, SF Bureau Chief at Forbes, interviewed Satya Nadella, Microsoft’s President of Server and Tools Business. The interview started with a general overview of the cloud, Nadella pointing out that we are in a middle of a major shift from a client-server programming model to a world of connected devices and continuous services. We need to redefine what an OS is, said Nadella, when you have a data center with 250,000 machines and 1M cores.

Reliability, security and availability are the main concerns customers have when considering moving to a cloud, according to Nadella. He draws attention to the fact that security in a public cloud is not more of an issue as it is in the private data center. The “hard shell” can be different, but the “soft core” is problematic because once someone has penetrated into the soft core, one can have the same security problems be it in a public cloud or a private infrastructure. “It all comes down to compliance like the one the enterprises are putting in place” and “having great encrypting technology”, “key management is a big issue because that can be compromised”, Nadella concluding that “security remains a big issue for the industry, but it is not only for the cloud but for any network.”

Without specifying any numbers, Nadella mentioned that many businesses move their websites to Windows Azure while still calling back to their private servers for identification and some of the data, while some businesses create applications from scratch.

Talking on the status of Windows Azure, Nadella remarked growing adoption of SaaS solutions, especially Office 365, saying the “50% of the fortune 2,000 who have used us are now using our online offerings,” and there are other customers who are interested in virtualization and are creating instances in the cloud for cost benefits. He mentioned that tens of thousands of customers have applied with Azure, some of them have started using it for development and testing, while others are creating their private clouds using Microsoft’s cloud solutions and Azure appliances.

Nadella rejected the idea that cloud computing is a threat for Microsoft, adding that this is just an inflection point as many others in their history when they had to adapt and come with new and innovative solutions that brought them in a leadership position. While cloud computing is meant to lower computing costs, and thus affecting Microsoft’s business, the trend today is an increase of using computing resources, which is helped by the cloud model, increase that in the end benefits Microsoft from a financial perspective.

As a related idea, Microsoft has just announced that starting July 1st all inbound traffic both peak and off-peak times will be free. It seems that Microsoft wants to attract more customers by lowering the entrance barrier so they can move their data onto Azure with less costs.

In New Directions for the Cloud panel, Lew Tucker (LT), VP and CTO of Cloud Computing at Cisco and Jon Weinman (JW), Worldwide Lead HP, discussed some of the trends in cloud computing for the next ten years.

LT said that they need to create network components that “expose the underline attributes of the network through virtualized abstractions so applications and cloud stacks can take advantage of that.”

LT: “Applications will become self-aware in terms of what resources they use.” He continued by explaining that elasticity has two sides: scaling-up followed by a scaling-down when the resources are not longer needed, so the application management software needs to take those resources into account to know exactly what are the costs and resource availability.

JW: “Hybrids are best under most circumstances.”

LT: “The whole debate between private and public cloud is going to go away, because cloud computing is about making pooled resources accessible in an elastic model.” Those who move their internal data centers to a cloud model are going to be successful if “they manage to run it as an Internal Service Provider”.

JW: “When you think enterprise resources coupled with cloud resources, besides that model of distant data centers, there are other models possible.” “If you have a multi-offer type of hosting/cloud provider, there are benefits of co-lo coupled with on demand resources being in the same physical facility.” Another architecture, that JW sees coming in the next 10 years is the one where “the network connecting enterprise resources to the cloud is a bus on a single chip.” He explains that such a move is needed for homomorphic encryption and cloud security.

Both Tucker and Weinman believe that the border between private and public clouds will slowly disappear, the results being hybrid clouds, because that’s the essence of the cloud: the ability to run applications on any available resource be it internal or external.

Hello stranger!

You need to Register an InfoQ account or to post comments. But there's so much more behind being registered.

Get the most out of the InfoQ experience.

Tell us what you think

Allowed html: a,b,br,blockquote,i,li,pre,u,ul,p

Email me replies to any of my messages in this thread

OpenCore CARS Vision by William Louth

Lew Tuckers vision "Applications will become self-aware in terms of what resources they use" seems to very similar to our own except for we see this not just confined to the cloud and the granularity of a resource much finer even business related if need be.

Cost Aware Runtimes & Services (CARS)
opencore.jinspired.com/?page_id=3063

OpenCore QoS for Apps Technology by William Louth

Lew Tucker "so the application management software needs to take those resources into account to know exactly what are the costs and resource availability."

Our view of resource availability and its reservation is a small as a inflight request even within low latency environments in which such things consume only a few hundred of microseconds of processing time.

QoS for Applications
opencore.jinspired.com/?page_id=3058

Allowed html: a,b,br,blockquote,i,li,pre,u,ul,p

Email me replies to any of my messages in this thread

Allowed html: a,b,br,blockquote,i,li,pre,u,ul,p

Email me replies to any of my messages in this thread

2 Discuss

Educational Content

General Feedback
Bugs
Advertising
Editorial
InfoQ.com and all content copyright © 2006-2013 C4Media Inc. InfoQ.com hosted at Contegix, the best ISP we've ever worked with.
Privacy policy
BT