Does Cloud Computing need Superfast Broadband?

At the recent PITCOM meeting on Cloud Computing one of the speakers was asked whether the availability of bandwidth would limit take up, including by small firms and teleworkers. The questioner was told this was not a serious issue. I have since been trying to find material on the bandwidth that would be needed by those seeking to cut their IT costs by transitioning to “The Cloud”. 

I have discovered that while costs are coming down, transmission charges are one of the main ways that Cloud Operators make their money and that pricing is in gigabytyes, which implies a lot  of data. I also discovered the problem of the “skinny straw” (through which a user accesses their Cloud providers) and the various approaches for reducing the amount of traffic that needs to flow through that bottleneck.

According to ZDnet the bandwidth demands generated by Cloud Computing are helping drive suppliers towards a 100 gigabit (yes gigabit not megabit) “sweet spot”. When PITCOM visited California in May 2000 the technology “sweet spot” was 10 megs: how the world moves on!

That scale of demand, plus the effect of “skinny straw” bottlenecks on its potential future revenue growth, appears to what is driving Google into becoming an infrastructure provider – at least at a pilot level . I doubt that, after all its lobbying on Net Neutrality, Google really wishes to more than demonstrate what can and should be done – ideally by others. 

However, a couple of days spent googling articles on the bandwidth that might be needed by those wishing to transition to the Cloud has left me little wiser as to the bandwidth actually needed by those who wish to transition their businesses to Cloud computing.

I started ringing round thsoe who I thought might be able to able to help but have only succeeded in homing in on some interesting questions: 

1) Does the effective use of Cloud Computing require access to bandwidth akin to the gigabit networks of a modern University campus? 

2) Are the 100 meg leased lines used by large organisations based in urban locations for their existing applications adequate? 

3) Can it be made to run over the “broadband” available to most UK consumers and SMEs: ranging from under 1 mbs to about 6 mbs with an average of 4 mbs download and 400 kbs upload, both falling by about 30% when the kids go on-line in the evening: Ofcom reveals the Uk’s real broadband speeds   

If the answer is 3), then Cloud Computing is clearly an idea whose time has come.

If the answer is 2), then its future outside the urban/corporate world of leased lines will depend on the speed at which “tier one” broadband: (100mbs +, symmetric and uncontended) is beginning to be rolled out in other parts of the world.

If it is 1), then it will only become commonplace when we have fibre to the SME or home: in jarogn terms –  FTTH, not FTTC or GPON.

These questions help put the targets in the Digital Britain report into context. So too does the Berkman Report on Next Generation Connectivity : commissioned by the FCC to help policy discussion in the USU.

It says “The speeds set out as future goals in the UK document as “very fast” are what would be considered as second tier speeds by the standards of what is available today in the best performing coutries: 40 – 50 Mbps download and 20 Mbps upload.”  

That report makes sobering reading. Even London does not make the OECD top 20 Cities for either download or upload speeds: unlike competitors like  Amsterdam, Berlin, Bern, Hamburg, Lyon, New York, Paris, Rotterdam, Tokyo or even Bratislava, Lisbon or Prague.

The Information Society Alliance – EURIM paper for those parliamentary candidates planning to use boradband access campaigns as part of their own election strategies was sent out for review last week. The immediate reaction from those representing some of the existing operators was that it referred to objectives that were unrealistically ambitious. The reaction from some of the candidates with a professional backbround in IT and Communications was that it was too modest.

This is clearly “an area of debate” – and I look forward to readers telling me how wrong I am – and giving me good references as to where I can find “reality”.



Join the conversation


Send me notifications when other members comment.

Please create a username to comment.


An interesting article. In answer to your questions, my thoughts based on expereince of using hosted services (e.g. SaaS) would suggest that you first need to consider what the 'Cloud' is doing. Is it simply providing consolidation of individual servers, with processors allocated as demand requires, or is it replacing functionality currently found on the users' PCs? The nature of the Cloud's role will affect the answer to your questions.

1) Does the effective use of Cloud Computing require access to bandwidth akin to the gigabit networks of a modern University campus? If the Cloud is primarily a file, database and webserver probably not, however if the Could is to run "Office"-type applications, e.g. spreadsheets and word-processors, then the bandwidth required will be significantly higher

2) Are the 100 meg leased lines used by large organisations based in urban locations for their existing applications adequate? That depends on the role of the Cloud and the answer to Question 1.

3) Can it be made to run over the "broadband" available to most UK consumers and SMEs? Probably not. As a heavy web users with a reasonably reliable 8Mbps copper-based downlink, I am very aware of the consequences of our current asymmetric broadband. Slow uploading of files and performance that is significantly below the headline figure in the marketing literature are experienced daily

As regard your question about 'good references as to where I can find "reality"' - there is some published material on broadband performance, but I have not seen any papers that specifically address the communications and networking bandwidths required for future applications including cloud computing. For that matter there appears to be only limited publicly available material on the bandwidths required to support current applications (browsing, games, etc).

1) It's all about the amount of data to send between client and servers in the cloud. Webbased "Office apps" don't send a lot of data so they can scale without lot of bandwidth.

2)The problem are not the productivity apps but highly bandwidth consuming sites like video-sharing.

3. Quality of Service (QoS) management between client and "the cloud" will become more important. But these technologies already exist. When dad and mum are doing their work from home , download speed for the children's pc's will be shrinked.