Visionary hardware systems:

Ubiquitous Computing (UC) has its origins in the development of service computing and cloud computing technologies. There are three key elements to providing the entire business IT strategy as a single service.

1. infinitely scalable hardware solutions without bottlenecks that can be re-deployed on the fly to meet the changing needs of the business.
2. a single copy of the data for the whole organisation that can be accessed speedily and seamlessly by all the different applications used in the business process – around the globe.
3. an understanding of the business process within the organisation that can be mapped across the UC model.

 

Business first model:

Ubiquitous Computing (UC) enables you to construct an IT solution for your business model rather than the having to orientate the business around your IT platform.

By adding flexibility to visionary technologies, the ubiquitous platform delivers freedom and efficiency for your business. The four key elements of traditional solutions are merged into a single, manageable and highly flexible IT infrastructure as described below:

1. Processing power:

The traditional model of discreet ‘boxes’ providing a single function within the global IT environment requires significantly more hardware than is really required. In providing a unique function, each box also demands redundancy for peak loading or failure.
Virtualisation partially fixes this problem; virtualisation allows you to create a pool of discreet systems with virtual servers that can process the applications in hand, whilst offering the flexibility of servicing other applications with any surplus processing capability.  It also offers the opportunity to move big applications in peak load to large servers or to smaller servers as the load reduces, though swapping systems between servers causes delay while services restart and if servers have different capabilities results can be unpredictable.
The main Achilles heal of virtualisation is physical resource; the mix and number of large/ small server systems available under real requirement is almost infinitely variable according to the call and load of applications and although process planning will give an indication of need, over-engineering will always occur as pressure points arise. The overhead required for system allocation is also not inconsiderable.

The UC model creates a fabric structure that separates the hardware from the virtual computer layer, providing a pool of processing available to the organisation as a whole and providing significant advantages:-

1. Flexibility of demand by applications can be accommodated in real time by simple resource allocation to existing servers, removing loading issues, delays, unpredictability of outcomes and individual server redundancy issues.
2. Total flexibility over the physical resource as servers can be structured as 1 core on a processor to 80 processors across multiple system boards. The concept of a server becomes a configuration file which can be simply changed as required.
3. As overall business growth occurs, adding additional hardware to the general processing pool delivers capability and flexibility to the whole organisation.
4. Lost processing resource from individual server failure is absorbed within the computing pool.

 

1. Networking:

Networking is the second key component of UC:
Networks associated with discreet systems often create bottlenecks, database applications are a prime example. If a database application exceeds the capacity of a single server, the intercommunication between the two servers can absorb up to 80% of the network traffic, leaving the balance for the application. Virtual servers can have the added complication of differing network response times.

The UC fabric computing model delivers significant advantages:

1.  a minimum interconnect between all servers of 40 Gbts per second, which closely maps the interconnect speed of SMP servers.
2. Interconnectivity and speed of the pool servers delivers a variable structure that can react quickly to need by redesigning it to mirror one large SMP server, many small independent servers or anything between the two options.
3. High speed interconnect between systems, the chassis and the outside world removes the need for a dedicated storage connection as all of the data can pass over the standard interconnect without significant impact on the computing power of the overall system.

 

1. Storage:

Storage focussing on the “Data is King” principle is the key function in the UC model:
The UC model holds the data in non weight status for the whole system, so that it can be redirected to anywhere within the system on an on call basis, enabling flexibility and speed across the system, even within the most dynamic IT model.
Data is stored across a number of storage devices that carry a variety of performance and overhead characteristics which ensures that the system has the appropriate data in the most convenient location.

Storage structure:
The ideal, but unfeasible solution of holding all data in memory would be the best solution. The UC model offers a strong compromise in which a multi-tier storage structure is adopted:

Tier 1:

1. Solid state storage holds current data enabling read access at close to memory speed and writes up to 70% of memory speed (provided it is constructed properly). This ensures that key data is available on demand across the system.
2. Mitigation of data loss in the event of power failure as solid state memory is persistent.

Tier 2:

1. Raid disc storage is utilised for less regularly accessed data. It provides good access speeds for medium storage and offers an ideal backup to tier 1 storage to reduce the risk of data loss.

Tier 3:

1. High capacity storage offers long term facilities with the capability to de-duplicate storage requirements and by adding compression provides the highest return on investment per Gbyte.

Although the above three storage methods are available in standard compute models, in combination within the UC model they provide direct access to a storage free server solution that fully implements network storage topologies with little or no latency. The result is that the data is stored in the most appropriate and efficient place for the business needs.

1. Applications:

Applications are the tools of the business, but also create some of the biggest issues:
The diverse range of software acquired from main stream, best of breed, or specialist vendors often employ differing technologies, methods and storage techniques as a result of individual company development strategy. To overcome the disparities, the modern IT solution has been designed to minimise the difficulties of intercommunication, but is focused at the lowest levels within the system.

The UC model delivers a very efficient solution:

1. All data is stored at a single source and in a single format.
2. A translator is employed to make the single source data available to all applications

 

The benefits from the UC model are significant:

1. Single copies of the data exist rather than multiple copies held on multiple machines resulting in massive synchronicity and data integrity problems.
2. Single source data ensures that users have access to the most up to date versions of files.
3. Storage capacity is significantly reduced.

 

Summary:
The Ubiquitous Computing model delivers major advantages by aligning the IT solution to the real business need, creating significant operational efficiencies and delivering savings on hardware, storage and management overheads:

The UC model provides a solution as powerful as the focused mainframe solutions from the past, but without the rigidity; a solution that has the flexibility of current interconnected systems, but without the intercommunication problems, duplication of hardware, software and data issues that can severely restrict business.

The UC model combines the best of both approaches in a highly flexible and powerful way, whist retaining the ability to scale (up and down) to directly mirror the needs of the business. With a lower infrastructure requirement the UC model can still accommodate any future growth by specifically targeting those elements that need to grow and simply linearly scaling the solution by adding hardware to the core building block.

Disruption is minimised: In current systems, if a key server fails, then the knock on effect throughout the organisation can be both serious. With the UC model, the loss of a machine does not have a material effect on the collective within the core building block. Any critical failure is immediately absorbed within the core and availability is maintained be resource reallocation.