A school's broadband connection is well and truly embedded into the everyday running of an establishment as well as providing them with a platform to deliver educational content for teaching and learning.
Schools are unique. Largely, they all start at the same time, all finish at the same time, have breaks and lunch time at the same time. Unlike corporate broadband, the usage isn’t smoothed by flexible working hours or work-from-home arrangements. Schools tend to use the internet in batches of thirty at a time; increasingly this is using one-to-one tablet devices all streaming the same YouTube video, for example. For us, and other ISPs, this means we have a very peaky traffic flow.
So, when the “internet goes down”, it is more than just an inconvenience – it impacts teacher confidence, disrupts children’s learning and can affect the smooth running in the school.
At RM Education, we have two key objectives for our Broadband Connectivity Service. The first is availability. We measure this closely, and monitor every school’s connection and proactively communicate with schools when it goes down to resume normal service as quickly as possible. The second objective is performance. We have a metric that measures how quickly traffic is flowing across our network to make sure that schools are receiving the performance they need – a highly available connection with poor performance is almost as bad as no connection at all.
Here are a few of the ways that we maintain high availability and performance across our Connectivity services for schools.
Plan for when hardware fails, not if
We double up on nearly everything in our network, so if half our network failed we’d still have enough bandwidth in the other half to service the needs of our customers without any performance degradation. We have two datacentres, two power feeds, hardware with dual power supplies using different power feeds, half our core network is in each datacentre location. Most of our services even include a level of resilience with our telecoms partner’s network too.
DDoS attacks are commonplace now. We’ve invested heavily in a service that monitors the traffic entering our network and automatically redirects traffic to a so called scrubbing centre to remove the DDoS traffic and allow normal traffic to flow without impacting schools. Typically we see over 100 attacks a year and because the DDoS protection takes effect within minutes, our schools are not impacted.
The key to maintaining great performance for schools is Capacity Management. RM Education is the fastest growing educationally focused ISP in the UK. In the last year, we’ve added over 800 schools to our service. In addition to these new customers, our forecasts show, and Google agrees with us, that we expect schools’ internet traffic to grow by 30% each year. So without really close capacity management, it can be very easy to get into a position where schools are seeing performance issues, particularly during peak times or when Microsoft or Apple release a significant update (or England play Wales in the Euros). And this is where we get into the detail a bit – it’s not just about having massive internet transit and peering capacity (we have about 100GBps of it), you have to be able to deliver that through your core network and out to customers. But a lot of hardware can get in the way. Internet filtering devices, load balancers, firewalls, virtual machine hosts, routers, switches, interconnects, fibre multiplexers all have capacity limitations, and without careful design it is easy for one of these to upset the service for many schools.
A key principle in capacity management is contention. Every ISP contends their core network. Selling an ‘uncontended leased line’ means the infrastructure into an ISP's network is not shared with anyone, in the same way a more consumer focused service might be, but once it is in the ISP's network the service is shared with everyone else. It’s no different to a school’s network. A school might have a 1Gbps core switch, but it will be providing services for 500 end devices each with 100Mbps network connections – that’s a total of 50Gbps, so a 50:1 contention ratio. Running an uncontended network is just not economic.
So contention is fine as long as it doesn’t turn into congestion. Once congested, a school’s service will suffer – buffering video, slow internet access, reduced access to key finance systems…the list goes on.
RM’s Connectivity Services team have been operating an Information Security Management System certified to ISO27001 for many years. Generally ISO27001 is seen as a security standard and it certainly does help us manage security aspects of our network, but InfoSec covers Confidentiality, Integrity and Availability (we include performance in this). As such we’ve developed systems and processes that allow us to monitor the here and now (we have 83 monitoring checks every second); analyse trends and plan for the future. Our great Engineering teams spend a significant proportion of their time planning out our capacity requirements and investments needed over the next 12 to 24 months. Our NOC (Network Operations Centre) resolve monitoring alerts and conduct maintenance duties 24 hours a day.
Our processes also accommodate the analysis of incidents that we experience. We’re not perfect and even with all the good stuff talked about here, issues will happen. Recovering them quickly (incident management) and learning from it (continuous improvement) is vital.
RM Education’s Connectivity Service operates to very high standards of availability and performance by designing in resilience, a continual focus on capacity; staying on top of increasing bandwidth requirements, all underpinned by externally certified processes and a fantastic team of people all focussed on the needs of our school customers.