Edge Applications in the Core network – saving us all money?

Posted on March 19, 2013


IBM and NSN have announced that they are putting app servers at the edge of the telco networks,  so what does this mean ?
This is a good thing: We are all using our phones and tablets more, and thus there is an ever increasing volume of data being handled over the mobile networks.
This is a bad thing: there is an ever increasing volume of data going over the mobile networks.

Why? you may well ask,  well it costs the telcos real cash money to transfer the data from the cell tower to the core network, and then on to the internet, and they tend to charge us in large tarrif steps based on the amount of data you use.  So this typically might be 0.5, 1, 4, and 8 GB of data, so you do need as an individual to use a lot more data before they get more money.  More fool them you might say.

It is also a problem because all this data leads to congestion, and slow networks.  The radio interface is variable enough without adding the apparently random effects of link congestion.  This leads to a poor user experience and frustrated users,  and this is a bad thing too.
There have been plans for many years to better use the computers in the cell towers to deliver some content and application support, because the best place to service a request is the nearest place to the device, and of course this is the cell tower.  This is not as easy as you might think,  because you might have been of the mind set that you could simply chuck a web cache in the tower, and it will serve the most popular content.  And this would work well, except for the
fact that for 2G and 3G the protocols used by the system at this point are not TCP/IP and so you cannot just use regular kit, but you also need an expensive and specialised piece of kit that can convert the protocol for you.
This will not be true for LTE as it is IP from end to end,  this will make it far easier to place regular solutions in the cell tower, so now we can place a web cache, a video playout facility, perhaps have standard services like self location lookup etc, running in the tower systems.
This save back hauling the requests to the main core network, and for content on towards the internet and the origin servers.
This is a good thing; it saves on latency (service responsiveness is dramatically improved), it reduces the amount of data retrieved over the core from the internet, and this reduces the congestion and all the effects of this. A large Tier 1 operator could probably save several hundred million € in reduced back haul from the cell towers to the core, and in the internet interconnects they have to maintain. This means they can charge less for data.  And you know what that is don’t you,  that’s right, a good thing.
So in summary put in a Content Delivery Network for streamed content, a Cache for HTTP content and local app servers (in the IMS model ?) for the
there is a downside,  and the biggee is that there are a _lot_ of cell towers,  here in the UK alone there are c. 23000 or them, so that’s a lot of locations to either deploy software too,  or more likely add equipment to.  Also it increases the amount of power and cooling required at each of these sites etc etc. which adds up when you have thousands of them,  so it is likely that the deployment would be to a selected set of the sites, where the impact is the greatest, eg with expensive backhaul (for example over microwave link) or where congestion is high (city centre).
Posted in: Architecture, mobile