Cloud tuning: getting the network up to speed
Get out of the slow lane
Many problems in IT can be put down to the simple but flawed assumption that resources are infinite.
It is obviously wrong but we repeatedly see examples of where it leads: running out of storage (the technology equivalent of the wardrobe principle that if you have the space, you will fill it), running too many applications at once, writing bloated code and trying to use network connections for purposes they were not designed for.
This last in particular has bitten the backsides of many an attempt to deliver and use cloud-based services. In the early days of selling online storage to consumers, for example, large quantities of space were offered that were virtually unusable.
Why are we waiting?
You don’t need a maths degree to work out that it could take a while to upload 5GB of information over an asymmetric (where uploads are much slower than downloads) 2Mb connection.
For corporates, bandwidth should not be such an issue, but home workers and smaller businesses frequently suffer the symptoms of inadequately specified or unreliable connections.
As well as bandwidth and reliability, latency can affect companies of all sizes. Response times from locally hosted applications are measurable in milliseconds, whereas we are all familiar with having to wait a second or five for web pages to load.
This may be not so bad when we are just surfing but it can become more than annoying when trying to access internet-based business applications – and indeed, it may break the terms of service level agreements.
Tolerance thresholds fall even lower when two computer systems are trying to interact (cloud-based or otherwise in hybrid environments.
Many technology experts are well aware of such issues, and they have beavered away for decades to resolve them.
For example we have HTTP, designed to enable asynchronous data transfers over less reliable connections. The accompanying RESTful protocols now offer the de facto way (rather than SOAP - Simple Object Access Protocol) for interfacing to web-based application services.
Other bright sparks are working on caching technologies for applications, video and other media streaming, voice protocols and so on, between a multiplicity of devices and across a widening variety of connection types.
Blame the internet
But today things are moving so fast that it becomes difficult to keep tabs on what problems are being solved where – within the network, by operating systems and management software, or in the application and delivery layers.
Solutions do not need to be complicated
In base terms, the internet will always run more slowly than local networking – that’s just electronics. There are smart people working on solutions, but organisations can recognise the real constraints inherent in the architecture and make decisions based on what is feasible and achievable.
Solutions do not need to be complicated. For example, a mobile application could enable access to a specific subset of application functions and data – logging, for example, the fact that a visit to the customer site has taken place and was a success.
Equally, cloud-based models may offer ways of doing things that we have never considered, enabling new services or new ways of working.
We certainly shouldn’t feel wedded to old models used for in-house applications. That path is more likely to lead to disappointment as a new set of constraints arises.
By abandoning the expectation that the internet should be an infinite resource, we can open opportunities to build new kinds of systems and services that make the best use of capabilities in the cloud as well as of any innovations we have achieved in-house. ®
Sponsored: IT evolution to a hybrid enterprise