Webupon

The Internet and Cloud Computing

Author: Peter Hartman

Today we are hearing a lot about cloud computing and what a wonderful revelation it is supposed to be. Let us clear up the doubts and inconsistencies and see whether or not it is for us.

The Cloud

In short; “the cloud” is a metaphor for the Internet derived from how the Internet has been long depicted in network diagrams such as Figure 1 below. The basic idea of the cloud is that users are able to connect to service providers and through them other users in a transparent manner without the knowledge of or even the need to know about the complexities of the underlying architecture and technologies that make the information technology services accessed and used by users possible.

Cloud Computing Services

Cloud style computing delivers IT-related capabilities “as a service”. Thus users are able to access information technology enabled services by way of the Internet (“the cloud”) without any expertise or knowledge needed on the user’s part.

Generally all control over; or administrative interaction with, the technologies and infrastructure required to deliver and support the “cloud services” on offer for end-user accessibility and the functionalities provided to the end-user by the said information technologies and infrastructure as “services in the cloud” are managed by “the cloud services” provider rather than by “the cloud services” clients. This is a double-edged sword with an upside and a downside that I will detail shortly but for now remember that not every “cloud” has a silver lining.

The Institute of Electronics and Electrical Engineers (IEEE) state that “Cloud computing is a paradigm in which information is permanently stored in servers on the Internet and cached temporarily on clients that include desktops, entertainment centers, table computers, notebooks, wall computers, handhelds, etc.”.

Cloud Computing Components

Think of cloud computing as a generally abstract concept incorporating such technologies as Software-as-a-Service (S-a-a-S), the whole plethora of technologies referred to under the umbrella of Web 2.0 functionalities and other mature and emerging technology trends that all rely heavily upon the Internet to deliver the communications channels between users and processing facilities and data centers in order to satisfy the computing needs of end-users.

A well known example of the “cloud computing” philosophy is Google Apps, which provides common business applications online that are accessed from a web browser, while the software and data are stored on Internet connected servers. You could say that in the strictest sense the Google search facility is a “cloud computing service”.

Cloud Computing and Google Search

Google uses automated bots called “spiders” to crawl the Internet. These spiders transmit information; back to Google’s data center, about the web pages that they have crawled. This information is then processed, classified and indexed by Google using a weighted relevance algorithm so that whenever a user requests information regarding any particular topic Google can look up their cached index database and return hopefully relevant links to the inquiring user.

Pre-Indexing – The reason that it all works so fast is that the scanning (crawling) and indexing of the Internet has already been done and all Google’s software need do is look up a referential database for associated links to the specified query parameters.

Performance – A big part of the Google speed magic is delivered by knowing that all they are required to deliver to the searcher is the first page of related topic headers, summaries and links. If the searcher requires more than they will be compiled and delivered to the user’s browser as and when requested.

The Laws of Processing – Using the basic processing law that “the information that you are most likely to need to process next is that which you have just processed” Google builds lists based on what they call page-rank.

The Cloud Factor – Where does cloud computing come into all of this you may well ask? The answer is that; as depicted in Figure 2 the Internet is used as the communications channel by a user to request a service (search) from a transparently Internet connected search service provider (Google) in real time.

• User sends search query request via the Internet to the transparently Internet connected search service provider (Alta Vista, Craig’s List, Yahoo or Google as in this case)
• The search service provider (Google in this case) performs some processing on the users behalf (searches their database index)
• Additional processing of the retrieved query results is performed by the search service provider to ensure that that the results are presented in a format that the user can use (browser and application specific)
• The results (the first page of relevant links) are then transmitted via the Internet (the cloud) to the user and the search results are stored in cache by the search service provider ready for additional access should that be required
• The user receives the results from the cloud and they are displayed in the format requested by the user (on-screen or printed etc). The user; if satisfied with the results returned may end the service at this point or if not yet satisfied request additional search query processing from the search service provider (Google once again) via the Internet.

So you see that using Google Search meets all the requirements necessary for this to be classified as “cloud computing”. As shown in Figure 2 above the Internet “cloud” is the communications medium and the requesting processes, signal transmission and routing, Google’s data center request processing, results compilation and search results transmission back to the requesting user via the Internet all take place transparently without the user’s knowledge or any control over the information technology systems involved in their successful search query. In a nutshell this is Cloud Computing.

Cloud Architectures and Infrastructures

Decentralized Models – Today we find that some successful cloud computing architectures have very little truly centralized infrastructure or billing systems. Examples here include Peer-to-Peer networks like BitTorrent and Skype along with “volunteer” computing implementations such as SETI@home.

Centralized Models – Other cloud computing implementations such as the Google search and Google Apps scenarios outlined above are patterned around a distributed centralized computing infrastructure.

Distributed Centralized Processing Model – Centralized because centrally located servers perform the processing and distributed since multiple processing facilities are incorporated into the production environment implementation.

The reasons for adopting a distributed centralized infrastructure are to provide for greater resiliency, scalability and redundancy of assets with the objective of eliminating potential single-point-of-failure scenarios.

This usually involves multiple ultra high-speed links to the Internet backbone for each such facility thereby reducing the impact of a single link failure bringing the entire cloud computing facility and its services down.

We must bear in mind that the majority of cloud computing services consist of reliable services delivered from state-of-the-art “green” data centers designed and built on compute (processing) and storage virtualization technologies. Not surprisingly blade centers and traditional rack mount systems in clusters play a leading role.

Anywhere Accessibility – Having services, applications and data available and accessible from anywhere and everywhere on the global stage with the Internet cloud acting as a single point of access for all of the computing needs of a group or many groups of large numbers of users is one of the primary benefits promised by the cloud computing model. Only time will tell as to whether or not this is delivered by cloud computing providers to the satisfaction of customers.

Commercial Cloud Computing Implementations

From a commercial perspective the cloud computing service provider; such as a web hosting service, will need to meet some very demanding and widely varied metrics requirements including Quality-of-Service (QoS) that will need specific and custom service level agreements (SLAs) far different to the majority of those we have seen in the past.

Open Standards – In these regards open standards (don’t forget that the Internet owes its success to being an open standard) and open source software have a lot to offer and may well prove to be the critical factors in the successful growth and deployment scenarios for the cloud computing world in the years ahead.

Reduced Capital Expenditure – Because customers of cloud computing service providers are not burdened with the fiscal outlays; that would normally accompany the rollout of new information technologies and their supporting infrastructure, they can budget their costs as a consumable service utility.

This means that they will pay for what they use in very much the same way as we pay for other utilities such as water, power and gas etc. There is also no need to overspend in order to make provision for peak and seasonal demands.

This is an area which costs business considerably as a recent Forester research report has shown that most companies use only 10% of their computing resources 95% of the time with public holidays and the Christmas period being the exceptions to standard usage patterns. With many businesses so reliant on this peak trade to ensure year-long viability they have no choice but to plan for the exception rather than the rule. “Make hay while the sun shines” is the modus operandi behind this type of strategy.

Utility and Subscription Billing Models – It is no surprise to find that many cloud computing offerings have indeed adopted the utility model and have tailored their services and billing structures accordingly. Yet another variation on this theme is the subscription model which charges customers on a subscription basis.

Both models have the own structural strengths and weaknesses and will probably co-exist side-by-side rather than directly compete head-to-head as it were. This latter prospect is given weight by the fact that some of the largest players in the cloud computing service provider game (Microsoft, IBM, HP, Google, Yahoo and Amazon.com) do in fact offer both billing structures to their clients.

Improved Efficiency and Cost Effectiveness – The sharing of computing resources among multiple “tenants” is one of the easiest to implement of all strategies to deliver greatly improved asset and resources utilization rates. This means that servers do what they were designed and built to do; that is to serve and have very little idle time. More bangs for your processing buck.

The resultant and significant savings in facility running costs are augmented by improvements in the speed of application development. Because customers do not have to engineer for peak loads dramatic improvements in computer capacity can be readily delivered.

Increased High-Speed Bandwidth – The central technologies that are making all this distributed centralized cloud computing a realistic and viable choice and hence its rapid uptake by business and consumers alike are the networking technologies that have recently delivered massive increases and capabilities at relatively minimal fiscal investments.

Improvements in fiber optic transmission media and 10 Gigabit Ethernet in combination deliver ultra high-speed bandwidth capabilities that make it possible to receive the same degree of agility and rapid response times from centralized infrastructure at all geographically dispersed sites.

It no longer matters which distributed centralized facility you connect with. The service response and delivery will be of the same high speed standard anywhere and everywhere. Nor does it matter from which of a client’s facilities the cloud computing services are accessed from.

This latter point seems to have had the most influence since branch offices can now have head office transmission and processing capabilities and performance. A fact not lost on the corporate world.

Cloud Computing Providers

Players in the cloud computing service provision game include the likes of Amazon, Google, Hewlett Packard, IBM, Intel, Microsoft, Salesforce, SAP and Yahoo!

Cloud Computing Users

Some of the early adopters of cloud computing services include individuals and small organizations through to very large conglomerates such as: General Electric, L’Oréal, Procter & Gamble, Valeo and the WWE.

It must be noted that traditional ISPs are not dragging the chain and are getting in on the cloud computing service provision industry in a big way. Tier 1 ISPs do after all have a huge advantage in that they already have, own or have peering agreements in place concerning the Internet backbone and IP Transit arrangements other would-be players will need to negotiate.

As already noted web hosting services are a major component of current cloud computing implementations and will undoubtedly grow exponentially as Web 2.0 and all its additional user capabilities become the true de facto standard for the Internet and websites in particular.

Reduced Control

It must be noted that the user of cloud computing services is for the large part totally dependent upon the cloud computing service provider for such administrative responsibilities as data security, backup, disaster planning and recovery, encryption, accessibility, authentication, regulation and many regulatory requirements. Data sanitization once data stored in a cloud computing environment has passed its user’s use by date is another concern.

Security Concerns

In general; the security aspects of a cloud computing service are primarily the concern and prerogative of the cloud computing service provider. All would-be customers should check very thoroughly the proposed providers intended operating procedures and practices as well as their track record in these regards.

It is very hard for the customer to verify the actual current security practices and implemented initiatives of a cloud computing service provider because the customer generally has no access to the provider’s facility which can be comprised of multiple facilities spread around the globe.

There can be little doubt that the Service Level Agreements (SLAs) that the customer of cloud computing services will need to negotiate with their prospective cloud computing services provider will for the reasons outlined above take on greater importance than has often been the case with Help Desk type SLAs in the past. Whether the cloud computing services provider develops a range of SLAs themselves or their prospective clients develop them remains to be seen.

What is clear is that the metrics of such SLAs do need to be very clearly detailed with clear and precise delineation of roles and responsibilities regarding a broad range of potential eventualities. It might pay to get the lawyers in on this one.