Cloud vs Dedicated Hosting
‘Cloud’ from Flickr ‘Web Hosting Servers’ from Flickr
Part 1: Cost The following trilogy of articles investigates the benefits and drawbacks of cloud hosting in comparison to the more traditional ‘all singing, all dancing’ hosting solutions of dedicated servers. It aims to discuss why enterprise consumers in particular are so tempted to migrate to the cloud as well as the barriers that often prevent them taking the leap.
Cost The first great migration to cloud computing was centered on its most obvious benefit in comparison to traditional computing, that of cost efficiencies. Cloud hosting, more specifically, is no different and offers a number of cost savings for enterprise which are missing from traditional dedicated platforms. Although dedicated servers can offer a good number of services which are much desired, in particular by enterprise consumers, the physical investment in them comes at a certain cost.
The Cost of Dedicated Servers A dedicated server may be completely ‘at the disposal’ of that one customer and, as such, may provide numerous security and performance benefits (as listed further on in series of posts) but maximum efficiency, in terms of the cost of the platforms versus the resource used, can only be realised if the platform is running at full capacity. If not, the consumer will inevitably be paying for capacity which they are getting any use from, because the cost of such platforms is met upfront. Essentially, customers are paying for physical capacity and it is then up to them to make use of it. Furthermore, if they need to increase resources (disk space, processing power etc) beyond current capacity, it requires upfront investment into the next ‘step up’, including the unused capacity that comes with that, as well as any reconfiguration and set up work that is required.
The Cost of Cloud Hosting Conversely, the cloud computing model, including cloud hosting, revolves around the concept of tapping into pooled computing resource on demand. In other words a consumer can access the resource they need as and when they need it, and, moreover, only pay for what they use. It can operate in a fashion akin to a household utility such as water or electricity where the consumer is plugged into the public service and is then charged for what they consume. The capacity of the shared resource is vast and so there is no stepping up from one fixed
© Stuart Mitchell 2013
Page 1 of 5
capacity to another and no additional setup costs therefore involved. In practice, if a business wants to try a new venture they need only invest in the resource they require whilst they require it, without taking on damaging longer term costs. What’s more, the costs incurred by the the maintenance of the underlying infrastructure (i.e., the pooled computing resource) can be diluted by economies of scale where there is no need for bespoke environments to be created for each consumer. This saving is perhaps less significant with cloud hosting on IaaS (Infrastructure as a Service) and PaaS (Platform as a Service) than other cloud services such as SaaS (Software as a Service) because there are less opportunities to standardise elements of the service, although it still returns sizeable savings compared to traditional localised computing environments. Finally, cloud platforms are less likely to involve lockins to long term contracts. This is largely caused by the fact that there isn’t the need for a cloud provider to invest so much upfront in the creation and configuration of the cloud platform and consequently seek a return on that investment over a fixed term. Without this need for set up each time a platform is provisioned, cloud hosting services, like other cloud services can be simply turned on or off for the customer as and when needed.
Part 2: Performance The second instalment of this series of posts looking at how cloud hosting platforms match up to traditional dedicated hosting platforms focusses on a number of issues which businesses look for in a package, including reliability, flexibility and responsiveness.
Timely For both the provider and the consumer there are considerable differences between cloud and dedicated in the time in which the services can be up and running. As mentioned previously, dedicated hosting platforms require considerable input from the provider, often in consultation with the client to understand the business requirements, quote for a service and then set up and provision that platform (one of the reasons for providers seeking a fixed term lock in to ensure return on that invested time). However, cloud hosting does not require the latter of these processes. Once the level of resources required by a business customer has been defined a quote can be calculated much quicker, because the resource is already in place, and the setup time is much reduced in some cases almost instant as the configuration needed to provide the final service is reduced. Consequently, dedicated hosting platforms may take weeks to provision whereas cloud hosting platforms can be provisioned in a matter of minutes or hours. What’s more, when extra resource is required from a cloud hosting platform this again can often be provided and scaled in no time at all without the need for further server setups regardless of the demand. Instead already configured pooled computer resource can be tapped into as it is required. Dedicated platforms on the other hand may require the installation of further hardware and solution stacks, for example, when the capacity of the existing setup has been exceeded, and so the time lag that that entails.
Reliability Each hosting platform has its own pluspoints in the area of reliability and, conversely, it own issues on which it may be bettered. Cloud computing fundamentally relies on the premise of pooled computing resource and so redundancy is built into the core model. Whether it be within a public or private cloud, the physical liability for the hosting platform will be spread across numerous servers and so the risk of hardware/solution stack issues causing downtime is greatly spread and reduced. If one server goes offline, the hosting service will be maintained without interruption on the remaining servers. What’s more, if the hosting service utilises services
© Stuart Mitchell 2013
Page 2 of 5
from disparate data centres it can also negate the risk of localised failures (e.g., power cuts) causing downtime. However, for a consumer the cloud model can involve ambiguities as to the stability and reliability of the underlying physical resources and the hosting provider themselves if they do not plump for a trusted provider with known network capabilities. The cloud moniker is easy and popular to apply to computing services without the necessary resources to ensure high reliability and performance and so the level of service experienced across providers can vary greatly. Dedicated platforms however, benefit from reduced risk of failures in the first place as the server’s resources are not shared with other business’s functions and therefore are not at risk from their potential security vulnerabilities (see Part 3) or from these functions draining shared resources (bandwidth, disk space etc). The flip side of this is that any failures may take the entire server offline, although dedicated services will employ back up systems (often tape back ups which are low maintenance) to ensure that functions can be restored as quickly as possible if failures do occur albeit not necessarily seamlessly as with cloud platforms.
Cloud Computing from Flickr
Part 3: Enterprise Focus The third post in this series looks at some of the pros and cons of dedicated and cloud hosting solutions when it comes to providing the services that enterprise customers actually demand. Much focus in the industry has in the past been concentrated on the technical capabilities of the respective platforms but the key to adoption across enterprise is how that technology sates business requirements.
Customer Experience A traditional weakness of cloud computing, and perhaps a consequence of its ondemand access model is the area of SLAs and targeting enterprise consumers’ needs effectively. The utility style of the service means that consumers have to some extent fitted their computing needs to the cloud services available rather than vice versa in order to benefit from the economies of scale and reduced costs. After all, the service is to a large extent defined and packaged up by the provider with the consumer tapping into it as and when they need it. Dedicated platforms have, in the past, outperformed cloud in this area with the ability to provide customisation and control over individual servers and the use of more suitable SLAs on better defines services. Businesses
© Stuart Mitchell 2013
Page 3 of 5
have been able to take their IT requirements to a provider of dedicated hosting and build the platform around it (cost permitting) leading to a more bespoke set up. However, there is now a concerted effort within the cloud sector to provide better targeted enterprise applications with inbuilt flexibility, scalability and security as well as SLAs which accurately reflect the performance of these services and the needs of enterprise. An example of this move away from a onesizefits all model is the development of the idea of cloud application stores where organisations can purchase the components they need individually to construct a cloud package which is tailored to their business needs. In other words, providers create and define individual components but customers configure their overall bespoke service using these elements.
Choice The benefits mentioned above and in the preceding posts in relation to the cloud result in arguably the key long term driver for enterprise adoption of cloud hosting and cloud computing in general, that of choice. Ultimately, the flexibility of the model means that anything is theoretically possible for an enterprise customer if they have the budget and their provider has the resources. The same can be said about traditional dedicated platforms (at a greater cost) but the scalability issues encountered by businesses using dedicated servers are, as stated, negated with cloud hosting by the removal of the concept of capacity. As mentioned previously dedicated platforms can be used to provide a bespoke hosting solution for enterprise customers at any level but once the platform is established any further changes to it may require time and expense. With cloud hosting, if a business wishes to try a particular project or campaign as a short term venture they can so with minimal lead time, payasyougo costing, and responsive scaling, thus reducing these costs and the resulting risks of the venture. Part 4 of this series of posts goes on to focus in more detail on the topic of security and how the two hosting solutions compare.
Part 4: Security Having compared cloud with traditional dedicated hosting solutions on their respective costs and performance issues in the preceding posts in this series, the final instalment provides further analysis of the two in regard to security issues.
Security For many private and enterprise customers, security is the primary area of concern when making the switch from traditional localised computing to cloud computing solutions, particularly when it comes to the topic of hosting. Businesses that require high levels of security to be applied to their hosting platforms have traditionally flocked to dedicated hosting solutions, to avoid the vulnerabilities introduced by sharing servers with other companies or business functions. These enterprise customers have since been somewhat reticent to make the switch to cloud (despite the efficiencies mentioned previously).
Dedicated Server Security Dedicated servers have, by design, features which are conducive to high levels of security in that they are individual platforms on discrete servers which are operated for single purposes i.e., they do not share disk space or computing power with other services or businesses. This distinction leads to a number of security benefits in terms of both protecting access to hosted data and the preservation of that data. To achieve these twin aims, the risk of hackers or malware accessing the data and/or corrupting it is minimised; by not having
© Stuart Mitchell 2013
Page 4 of 5
any other functions/companies sharing the hosting platform it reduces the number of possible points of entry/access and therefore the number of security vulnerabilities on the server. What’s more, a business sharing a host server would have no control over the effectiveness of the measures taken to secure these vulnerabilities if they are sharing the server with third party businesses. The dedicated model also removes the competing demands placed on the physical computing capabilities of the server by other hosting platforms/solutions stacks/businesses’ IT projects, meaning that there is less risk of server or network failures leading to the unavailability or loss of data.
Cloud Hosting Security Cloud Hosting platforms therefore need to readdress these issues as they fundamentally rely on the concept of shared or pooled computing resource. Public cloud models will struggle to offer the same protection as a dedicated platform because they not only share physical hosting infrastructure across multiple virtualised hosting platforms for disparate customers, but have further vulnerabilities in that the access points to such services are across public networks in other words anyone can ‘knock on the door’ and any information being transferred between access point and server is at risk of being intercepted. Furthermore, one organisation who is a consumer of the service has no influence or control over the trustworthiness of others who may have signed up to share these pooled resources. The answer to dedicated platforms for cloud computing is the private cloud. This model relies on the concept of ringfencing a pool of computing resources for the use of a single organisation to eliminate the vulnerabilities of sharing. The concept has a variety of ways in which it can be physically implemented but where it involves a physically distinct pool of servers it can remove the aforementioned risks of sharing with third parties. In addition the use of a physically distinct line for access or onsite location of the servers can negate the risks of data being intercepted in transit or of unwanted access to the platform. However, by implementing measures such as these, organisations eliminate many of the economies of scale that make the cloud so attractive in the first place. Consequently, private clouds are often created using virtualisation to create ring fenced virtual networks of servers and secured access to those with technologies such as MPLS and VPN. These virtualised private clouds are becoming more and more secure and whilst they may not quite rival the physical independence of dedicated servers of localised private clouds there is a determination in the industry to close the gap and allow enterprise to benefit from the cost efficiencies and scalability benefits of cloud hosting without compromising on their security.
© Stuart Mitchell 2013
Page 5 of 5